Commit | Line | Data |
---|---|---|
f46b5a66 CH |
1 | /* |
2 | * Copyright (C) 2007 Oracle. All rights reserved. | |
3 | * | |
4 | * This program is free software; you can redistribute it and/or | |
5 | * modify it under the terms of the GNU General Public | |
6 | * License v2 as published by the Free Software Foundation. | |
7 | * | |
8 | * This program is distributed in the hope that it will be useful, | |
9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
11 | * General Public License for more details. | |
12 | * | |
13 | * You should have received a copy of the GNU General Public | |
14 | * License along with this program; if not, write to the | |
15 | * Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
16 | * Boston, MA 021110-1307, USA. | |
17 | */ | |
18 | ||
19 | #include <linux/kernel.h> | |
20 | #include <linux/bio.h> | |
21 | #include <linux/buffer_head.h> | |
22 | #include <linux/file.h> | |
23 | #include <linux/fs.h> | |
cb8e7090 | 24 | #include <linux/fsnotify.h> |
f46b5a66 CH |
25 | #include <linux/pagemap.h> |
26 | #include <linux/highmem.h> | |
27 | #include <linux/time.h> | |
28 | #include <linux/init.h> | |
29 | #include <linux/string.h> | |
f46b5a66 | 30 | #include <linux/backing-dev.h> |
cb8e7090 | 31 | #include <linux/mount.h> |
f46b5a66 | 32 | #include <linux/mpage.h> |
cb8e7090 | 33 | #include <linux/namei.h> |
f46b5a66 CH |
34 | #include <linux/swap.h> |
35 | #include <linux/writeback.h> | |
36 | #include <linux/statfs.h> | |
37 | #include <linux/compat.h> | |
38 | #include <linux/bit_spinlock.h> | |
cb8e7090 | 39 | #include <linux/security.h> |
f46b5a66 | 40 | #include <linux/xattr.h> |
7ea394f1 | 41 | #include <linux/vmalloc.h> |
5a0e3ad6 | 42 | #include <linux/slab.h> |
f7039b1d | 43 | #include <linux/blkdev.h> |
4b4e25f2 | 44 | #include "compat.h" |
f46b5a66 CH |
45 | #include "ctree.h" |
46 | #include "disk-io.h" | |
47 | #include "transaction.h" | |
48 | #include "btrfs_inode.h" | |
49 | #include "ioctl.h" | |
50 | #include "print-tree.h" | |
51 | #include "volumes.h" | |
925baedd | 52 | #include "locking.h" |
581bb050 | 53 | #include "inode-map.h" |
f46b5a66 | 54 | |
6cbff00f CH |
55 | /* Mask out flags that are inappropriate for the given type of inode. */ |
56 | static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags) | |
57 | { | |
58 | if (S_ISDIR(mode)) | |
59 | return flags; | |
60 | else if (S_ISREG(mode)) | |
61 | return flags & ~FS_DIRSYNC_FL; | |
62 | else | |
63 | return flags & (FS_NODUMP_FL | FS_NOATIME_FL); | |
64 | } | |
65 | ||
66 | /* | |
67 | * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl. | |
68 | */ | |
69 | static unsigned int btrfs_flags_to_ioctl(unsigned int flags) | |
70 | { | |
71 | unsigned int iflags = 0; | |
72 | ||
73 | if (flags & BTRFS_INODE_SYNC) | |
74 | iflags |= FS_SYNC_FL; | |
75 | if (flags & BTRFS_INODE_IMMUTABLE) | |
76 | iflags |= FS_IMMUTABLE_FL; | |
77 | if (flags & BTRFS_INODE_APPEND) | |
78 | iflags |= FS_APPEND_FL; | |
79 | if (flags & BTRFS_INODE_NODUMP) | |
80 | iflags |= FS_NODUMP_FL; | |
81 | if (flags & BTRFS_INODE_NOATIME) | |
82 | iflags |= FS_NOATIME_FL; | |
83 | if (flags & BTRFS_INODE_DIRSYNC) | |
84 | iflags |= FS_DIRSYNC_FL; | |
d0092bdd LZ |
85 | if (flags & BTRFS_INODE_NODATACOW) |
86 | iflags |= FS_NOCOW_FL; | |
87 | ||
88 | if ((flags & BTRFS_INODE_COMPRESS) && !(flags & BTRFS_INODE_NOCOMPRESS)) | |
89 | iflags |= FS_COMPR_FL; | |
90 | else if (flags & BTRFS_INODE_NOCOMPRESS) | |
91 | iflags |= FS_NOCOMP_FL; | |
6cbff00f CH |
92 | |
93 | return iflags; | |
94 | } | |
95 | ||
96 | /* | |
97 | * Update inode->i_flags based on the btrfs internal flags. | |
98 | */ | |
99 | void btrfs_update_iflags(struct inode *inode) | |
100 | { | |
101 | struct btrfs_inode *ip = BTRFS_I(inode); | |
102 | ||
103 | inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC); | |
104 | ||
105 | if (ip->flags & BTRFS_INODE_SYNC) | |
106 | inode->i_flags |= S_SYNC; | |
107 | if (ip->flags & BTRFS_INODE_IMMUTABLE) | |
108 | inode->i_flags |= S_IMMUTABLE; | |
109 | if (ip->flags & BTRFS_INODE_APPEND) | |
110 | inode->i_flags |= S_APPEND; | |
111 | if (ip->flags & BTRFS_INODE_NOATIME) | |
112 | inode->i_flags |= S_NOATIME; | |
113 | if (ip->flags & BTRFS_INODE_DIRSYNC) | |
114 | inode->i_flags |= S_DIRSYNC; | |
115 | } | |
116 | ||
117 | /* | |
118 | * Inherit flags from the parent inode. | |
119 | * | |
120 | * Unlike extN we don't have any flags we don't want to inherit currently. | |
121 | */ | |
122 | void btrfs_inherit_iflags(struct inode *inode, struct inode *dir) | |
123 | { | |
0b4dcea5 CM |
124 | unsigned int flags; |
125 | ||
126 | if (!dir) | |
127 | return; | |
128 | ||
129 | flags = BTRFS_I(dir)->flags; | |
6cbff00f CH |
130 | |
131 | if (S_ISREG(inode->i_mode)) | |
132 | flags &= ~BTRFS_INODE_DIRSYNC; | |
133 | else if (!S_ISDIR(inode->i_mode)) | |
134 | flags &= (BTRFS_INODE_NODUMP | BTRFS_INODE_NOATIME); | |
135 | ||
136 | BTRFS_I(inode)->flags = flags; | |
137 | btrfs_update_iflags(inode); | |
138 | } | |
139 | ||
140 | static int btrfs_ioctl_getflags(struct file *file, void __user *arg) | |
141 | { | |
142 | struct btrfs_inode *ip = BTRFS_I(file->f_path.dentry->d_inode); | |
143 | unsigned int flags = btrfs_flags_to_ioctl(ip->flags); | |
144 | ||
145 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
146 | return -EFAULT; | |
147 | return 0; | |
148 | } | |
149 | ||
75e7cb7f LB |
150 | static int check_flags(unsigned int flags) |
151 | { | |
152 | if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \ | |
153 | FS_NOATIME_FL | FS_NODUMP_FL | \ | |
154 | FS_SYNC_FL | FS_DIRSYNC_FL | \ | |
e1e8fb6a LZ |
155 | FS_NOCOMP_FL | FS_COMPR_FL | |
156 | FS_NOCOW_FL)) | |
75e7cb7f LB |
157 | return -EOPNOTSUPP; |
158 | ||
159 | if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL)) | |
160 | return -EINVAL; | |
161 | ||
75e7cb7f LB |
162 | return 0; |
163 | } | |
164 | ||
6cbff00f CH |
165 | static int btrfs_ioctl_setflags(struct file *file, void __user *arg) |
166 | { | |
167 | struct inode *inode = file->f_path.dentry->d_inode; | |
168 | struct btrfs_inode *ip = BTRFS_I(inode); | |
169 | struct btrfs_root *root = ip->root; | |
170 | struct btrfs_trans_handle *trans; | |
171 | unsigned int flags, oldflags; | |
172 | int ret; | |
173 | ||
b83cc969 LZ |
174 | if (btrfs_root_readonly(root)) |
175 | return -EROFS; | |
176 | ||
6cbff00f CH |
177 | if (copy_from_user(&flags, arg, sizeof(flags))) |
178 | return -EFAULT; | |
179 | ||
75e7cb7f LB |
180 | ret = check_flags(flags); |
181 | if (ret) | |
182 | return ret; | |
f46b5a66 | 183 | |
2e149670 | 184 | if (!inode_owner_or_capable(inode)) |
6cbff00f CH |
185 | return -EACCES; |
186 | ||
187 | mutex_lock(&inode->i_mutex); | |
188 | ||
189 | flags = btrfs_mask_flags(inode->i_mode, flags); | |
190 | oldflags = btrfs_flags_to_ioctl(ip->flags); | |
191 | if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) { | |
192 | if (!capable(CAP_LINUX_IMMUTABLE)) { | |
193 | ret = -EPERM; | |
194 | goto out_unlock; | |
195 | } | |
196 | } | |
197 | ||
198 | ret = mnt_want_write(file->f_path.mnt); | |
199 | if (ret) | |
200 | goto out_unlock; | |
201 | ||
202 | if (flags & FS_SYNC_FL) | |
203 | ip->flags |= BTRFS_INODE_SYNC; | |
204 | else | |
205 | ip->flags &= ~BTRFS_INODE_SYNC; | |
206 | if (flags & FS_IMMUTABLE_FL) | |
207 | ip->flags |= BTRFS_INODE_IMMUTABLE; | |
208 | else | |
209 | ip->flags &= ~BTRFS_INODE_IMMUTABLE; | |
210 | if (flags & FS_APPEND_FL) | |
211 | ip->flags |= BTRFS_INODE_APPEND; | |
212 | else | |
213 | ip->flags &= ~BTRFS_INODE_APPEND; | |
214 | if (flags & FS_NODUMP_FL) | |
215 | ip->flags |= BTRFS_INODE_NODUMP; | |
216 | else | |
217 | ip->flags &= ~BTRFS_INODE_NODUMP; | |
218 | if (flags & FS_NOATIME_FL) | |
219 | ip->flags |= BTRFS_INODE_NOATIME; | |
220 | else | |
221 | ip->flags &= ~BTRFS_INODE_NOATIME; | |
222 | if (flags & FS_DIRSYNC_FL) | |
223 | ip->flags |= BTRFS_INODE_DIRSYNC; | |
224 | else | |
225 | ip->flags &= ~BTRFS_INODE_DIRSYNC; | |
e1e8fb6a LZ |
226 | if (flags & FS_NOCOW_FL) |
227 | ip->flags |= BTRFS_INODE_NODATACOW; | |
228 | else | |
229 | ip->flags &= ~BTRFS_INODE_NODATACOW; | |
6cbff00f | 230 | |
75e7cb7f LB |
231 | /* |
232 | * The COMPRESS flag can only be changed by users, while the NOCOMPRESS | |
233 | * flag may be changed automatically if compression code won't make | |
234 | * things smaller. | |
235 | */ | |
236 | if (flags & FS_NOCOMP_FL) { | |
237 | ip->flags &= ~BTRFS_INODE_COMPRESS; | |
238 | ip->flags |= BTRFS_INODE_NOCOMPRESS; | |
239 | } else if (flags & FS_COMPR_FL) { | |
240 | ip->flags |= BTRFS_INODE_COMPRESS; | |
241 | ip->flags &= ~BTRFS_INODE_NOCOMPRESS; | |
ebcb904d LZ |
242 | } else { |
243 | ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS); | |
75e7cb7f | 244 | } |
6cbff00f CH |
245 | |
246 | trans = btrfs_join_transaction(root, 1); | |
3612b495 | 247 | BUG_ON(IS_ERR(trans)); |
6cbff00f CH |
248 | |
249 | ret = btrfs_update_inode(trans, root, inode); | |
250 | BUG_ON(ret); | |
251 | ||
252 | btrfs_update_iflags(inode); | |
253 | inode->i_ctime = CURRENT_TIME; | |
254 | btrfs_end_transaction(trans, root); | |
255 | ||
256 | mnt_drop_write(file->f_path.mnt); | |
2d4e6f6a | 257 | |
258 | ret = 0; | |
6cbff00f CH |
259 | out_unlock: |
260 | mutex_unlock(&inode->i_mutex); | |
2d4e6f6a | 261 | return ret; |
6cbff00f CH |
262 | } |
263 | ||
264 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) | |
265 | { | |
266 | struct inode *inode = file->f_path.dentry->d_inode; | |
267 | ||
268 | return put_user(inode->i_generation, arg); | |
269 | } | |
f46b5a66 | 270 | |
f7039b1d LD |
271 | static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg) |
272 | { | |
273 | struct btrfs_root *root = fdentry(file)->d_sb->s_fs_info; | |
274 | struct btrfs_fs_info *fs_info = root->fs_info; | |
275 | struct btrfs_device *device; | |
276 | struct request_queue *q; | |
277 | struct fstrim_range range; | |
278 | u64 minlen = ULLONG_MAX; | |
279 | u64 num_devices = 0; | |
280 | int ret; | |
281 | ||
282 | if (!capable(CAP_SYS_ADMIN)) | |
283 | return -EPERM; | |
284 | ||
1f78160c XG |
285 | rcu_read_lock(); |
286 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
287 | dev_list) { | |
f7039b1d LD |
288 | if (!device->bdev) |
289 | continue; | |
290 | q = bdev_get_queue(device->bdev); | |
291 | if (blk_queue_discard(q)) { | |
292 | num_devices++; | |
293 | minlen = min((u64)q->limits.discard_granularity, | |
294 | minlen); | |
295 | } | |
296 | } | |
1f78160c | 297 | rcu_read_unlock(); |
f7039b1d LD |
298 | if (!num_devices) |
299 | return -EOPNOTSUPP; | |
300 | ||
301 | if (copy_from_user(&range, arg, sizeof(range))) | |
302 | return -EFAULT; | |
303 | ||
304 | range.minlen = max(range.minlen, minlen); | |
305 | ret = btrfs_trim_fs(root, &range); | |
306 | if (ret < 0) | |
307 | return ret; | |
308 | ||
309 | if (copy_to_user(arg, &range, sizeof(range))) | |
310 | return -EFAULT; | |
311 | ||
312 | return 0; | |
313 | } | |
314 | ||
cb8e7090 CH |
315 | static noinline int create_subvol(struct btrfs_root *root, |
316 | struct dentry *dentry, | |
72fd032e SW |
317 | char *name, int namelen, |
318 | u64 *async_transid) | |
f46b5a66 CH |
319 | { |
320 | struct btrfs_trans_handle *trans; | |
321 | struct btrfs_key key; | |
322 | struct btrfs_root_item root_item; | |
323 | struct btrfs_inode_item *inode_item; | |
324 | struct extent_buffer *leaf; | |
76dda93c | 325 | struct btrfs_root *new_root; |
6a912213 JB |
326 | struct dentry *parent = dget_parent(dentry); |
327 | struct inode *dir; | |
f46b5a66 CH |
328 | int ret; |
329 | int err; | |
330 | u64 objectid; | |
331 | u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; | |
3de4586c | 332 | u64 index = 0; |
f46b5a66 | 333 | |
581bb050 | 334 | ret = btrfs_find_free_objectid(root->fs_info->tree_root, &objectid); |
6a912213 JB |
335 | if (ret) { |
336 | dput(parent); | |
a22285a6 | 337 | return ret; |
6a912213 JB |
338 | } |
339 | ||
340 | dir = parent->d_inode; | |
341 | ||
9ed74f2d JB |
342 | /* |
343 | * 1 - inode item | |
344 | * 2 - refs | |
345 | * 1 - root item | |
346 | * 2 - dir items | |
347 | */ | |
a22285a6 | 348 | trans = btrfs_start_transaction(root, 6); |
6a912213 JB |
349 | if (IS_ERR(trans)) { |
350 | dput(parent); | |
a22285a6 | 351 | return PTR_ERR(trans); |
6a912213 | 352 | } |
f46b5a66 | 353 | |
5d4f98a2 YZ |
354 | leaf = btrfs_alloc_free_block(trans, root, root->leafsize, |
355 | 0, objectid, NULL, 0, 0, 0); | |
8e8a1e31 JB |
356 | if (IS_ERR(leaf)) { |
357 | ret = PTR_ERR(leaf); | |
358 | goto fail; | |
359 | } | |
f46b5a66 | 360 | |
5d4f98a2 | 361 | memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header)); |
f46b5a66 CH |
362 | btrfs_set_header_bytenr(leaf, leaf->start); |
363 | btrfs_set_header_generation(leaf, trans->transid); | |
5d4f98a2 | 364 | btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV); |
f46b5a66 CH |
365 | btrfs_set_header_owner(leaf, objectid); |
366 | ||
367 | write_extent_buffer(leaf, root->fs_info->fsid, | |
368 | (unsigned long)btrfs_header_fsid(leaf), | |
369 | BTRFS_FSID_SIZE); | |
5d4f98a2 YZ |
370 | write_extent_buffer(leaf, root->fs_info->chunk_tree_uuid, |
371 | (unsigned long)btrfs_header_chunk_tree_uuid(leaf), | |
372 | BTRFS_UUID_SIZE); | |
f46b5a66 CH |
373 | btrfs_mark_buffer_dirty(leaf); |
374 | ||
375 | inode_item = &root_item.inode; | |
376 | memset(inode_item, 0, sizeof(*inode_item)); | |
377 | inode_item->generation = cpu_to_le64(1); | |
378 | inode_item->size = cpu_to_le64(3); | |
379 | inode_item->nlink = cpu_to_le32(1); | |
a76a3cd4 | 380 | inode_item->nbytes = cpu_to_le64(root->leafsize); |
f46b5a66 CH |
381 | inode_item->mode = cpu_to_le32(S_IFDIR | 0755); |
382 | ||
08fe4db1 LZ |
383 | root_item.flags = 0; |
384 | root_item.byte_limit = 0; | |
385 | inode_item->flags = cpu_to_le64(BTRFS_INODE_ROOT_ITEM_INIT); | |
386 | ||
f46b5a66 | 387 | btrfs_set_root_bytenr(&root_item, leaf->start); |
84234f3a | 388 | btrfs_set_root_generation(&root_item, trans->transid); |
f46b5a66 CH |
389 | btrfs_set_root_level(&root_item, 0); |
390 | btrfs_set_root_refs(&root_item, 1); | |
86b9f2ec | 391 | btrfs_set_root_used(&root_item, leaf->len); |
80ff3856 | 392 | btrfs_set_root_last_snapshot(&root_item, 0); |
f46b5a66 CH |
393 | |
394 | memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress)); | |
395 | root_item.drop_level = 0; | |
396 | ||
925baedd | 397 | btrfs_tree_unlock(leaf); |
f46b5a66 CH |
398 | free_extent_buffer(leaf); |
399 | leaf = NULL; | |
400 | ||
401 | btrfs_set_root_dirid(&root_item, new_dirid); | |
402 | ||
403 | key.objectid = objectid; | |
5d4f98a2 | 404 | key.offset = 0; |
f46b5a66 CH |
405 | btrfs_set_key_type(&key, BTRFS_ROOT_ITEM_KEY); |
406 | ret = btrfs_insert_root(trans, root->fs_info->tree_root, &key, | |
407 | &root_item); | |
408 | if (ret) | |
409 | goto fail; | |
410 | ||
76dda93c YZ |
411 | key.offset = (u64)-1; |
412 | new_root = btrfs_read_fs_root_no_name(root->fs_info, &key); | |
413 | BUG_ON(IS_ERR(new_root)); | |
414 | ||
415 | btrfs_record_root_in_trans(trans, new_root); | |
416 | ||
417 | ret = btrfs_create_subvol_root(trans, new_root, new_dirid, | |
418 | BTRFS_I(dir)->block_group); | |
f46b5a66 CH |
419 | /* |
420 | * insert the directory item | |
421 | */ | |
3de4586c CM |
422 | ret = btrfs_set_inode_index(dir, &index); |
423 | BUG_ON(ret); | |
424 | ||
425 | ret = btrfs_insert_dir_item(trans, root, | |
16cdcec7 | 426 | name, namelen, dir, &key, |
3de4586c | 427 | BTRFS_FT_DIR, index); |
f46b5a66 CH |
428 | if (ret) |
429 | goto fail; | |
0660b5af | 430 | |
52c26179 YZ |
431 | btrfs_i_size_write(dir, dir->i_size + namelen * 2); |
432 | ret = btrfs_update_inode(trans, root, dir); | |
433 | BUG_ON(ret); | |
434 | ||
0660b5af | 435 | ret = btrfs_add_root_ref(trans, root->fs_info->tree_root, |
4df27c4d | 436 | objectid, root->root_key.objectid, |
33345d01 | 437 | btrfs_ino(dir), index, name, namelen); |
0660b5af | 438 | |
76dda93c | 439 | BUG_ON(ret); |
f46b5a66 | 440 | |
76dda93c | 441 | d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry)); |
f46b5a66 | 442 | fail: |
6a912213 | 443 | dput(parent); |
72fd032e SW |
444 | if (async_transid) { |
445 | *async_transid = trans->transid; | |
446 | err = btrfs_commit_transaction_async(trans, root, 1); | |
447 | } else { | |
448 | err = btrfs_commit_transaction(trans, root); | |
449 | } | |
f46b5a66 CH |
450 | if (err && !ret) |
451 | ret = err; | |
f46b5a66 CH |
452 | return ret; |
453 | } | |
454 | ||
72fd032e | 455 | static int create_snapshot(struct btrfs_root *root, struct dentry *dentry, |
b83cc969 LZ |
456 | char *name, int namelen, u64 *async_transid, |
457 | bool readonly) | |
f46b5a66 | 458 | { |
2e4bfab9 | 459 | struct inode *inode; |
6a912213 | 460 | struct dentry *parent; |
f46b5a66 CH |
461 | struct btrfs_pending_snapshot *pending_snapshot; |
462 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 463 | int ret; |
f46b5a66 CH |
464 | |
465 | if (!root->ref_cows) | |
466 | return -EINVAL; | |
467 | ||
3de4586c | 468 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS); |
a22285a6 YZ |
469 | if (!pending_snapshot) |
470 | return -ENOMEM; | |
471 | ||
472 | btrfs_init_block_rsv(&pending_snapshot->block_rsv); | |
3de4586c | 473 | pending_snapshot->dentry = dentry; |
f46b5a66 | 474 | pending_snapshot->root = root; |
b83cc969 | 475 | pending_snapshot->readonly = readonly; |
a22285a6 YZ |
476 | |
477 | trans = btrfs_start_transaction(root->fs_info->extent_root, 5); | |
478 | if (IS_ERR(trans)) { | |
479 | ret = PTR_ERR(trans); | |
480 | goto fail; | |
481 | } | |
482 | ||
483 | ret = btrfs_snap_reserve_metadata(trans, pending_snapshot); | |
484 | BUG_ON(ret); | |
485 | ||
f46b5a66 CH |
486 | list_add(&pending_snapshot->list, |
487 | &trans->transaction->pending_snapshots); | |
72fd032e SW |
488 | if (async_transid) { |
489 | *async_transid = trans->transid; | |
490 | ret = btrfs_commit_transaction_async(trans, | |
491 | root->fs_info->extent_root, 1); | |
492 | } else { | |
493 | ret = btrfs_commit_transaction(trans, | |
494 | root->fs_info->extent_root); | |
495 | } | |
2e4bfab9 | 496 | BUG_ON(ret); |
a22285a6 YZ |
497 | |
498 | ret = pending_snapshot->error; | |
499 | if (ret) | |
500 | goto fail; | |
501 | ||
66b4ffd1 JB |
502 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
503 | if (ret) | |
504 | goto fail; | |
f46b5a66 | 505 | |
6a912213 JB |
506 | parent = dget_parent(dentry); |
507 | inode = btrfs_lookup_dentry(parent->d_inode, dentry); | |
508 | dput(parent); | |
2e4bfab9 YZ |
509 | if (IS_ERR(inode)) { |
510 | ret = PTR_ERR(inode); | |
511 | goto fail; | |
512 | } | |
513 | BUG_ON(!inode); | |
514 | d_instantiate(dentry, inode); | |
515 | ret = 0; | |
516 | fail: | |
a22285a6 | 517 | kfree(pending_snapshot); |
f46b5a66 CH |
518 | return ret; |
519 | } | |
520 | ||
4260f7c7 SW |
521 | /* copy of check_sticky in fs/namei.c() |
522 | * It's inline, so penalty for filesystems that don't use sticky bit is | |
523 | * minimal. | |
524 | */ | |
525 | static inline int btrfs_check_sticky(struct inode *dir, struct inode *inode) | |
526 | { | |
527 | uid_t fsuid = current_fsuid(); | |
528 | ||
529 | if (!(dir->i_mode & S_ISVTX)) | |
530 | return 0; | |
531 | if (inode->i_uid == fsuid) | |
532 | return 0; | |
533 | if (dir->i_uid == fsuid) | |
534 | return 0; | |
535 | return !capable(CAP_FOWNER); | |
536 | } | |
537 | ||
538 | /* copy of may_delete in fs/namei.c() | |
539 | * Check whether we can remove a link victim from directory dir, check | |
540 | * whether the type of victim is right. | |
541 | * 1. We can't do it if dir is read-only (done in permission()) | |
542 | * 2. We should have write and exec permissions on dir | |
543 | * 3. We can't remove anything from append-only dir | |
544 | * 4. We can't do anything with immutable dir (done in permission()) | |
545 | * 5. If the sticky bit on dir is set we should either | |
546 | * a. be owner of dir, or | |
547 | * b. be owner of victim, or | |
548 | * c. have CAP_FOWNER capability | |
549 | * 6. If the victim is append-only or immutable we can't do antyhing with | |
550 | * links pointing to it. | |
551 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
552 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
553 | * 9. We can't remove a root or mountpoint. | |
554 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
555 | * nfs_async_unlink(). | |
556 | */ | |
557 | ||
558 | static int btrfs_may_delete(struct inode *dir,struct dentry *victim,int isdir) | |
559 | { | |
560 | int error; | |
561 | ||
562 | if (!victim->d_inode) | |
563 | return -ENOENT; | |
564 | ||
565 | BUG_ON(victim->d_parent->d_inode != dir); | |
566 | audit_inode_child(victim, dir); | |
567 | ||
568 | error = inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
569 | if (error) | |
570 | return error; | |
571 | if (IS_APPEND(dir)) | |
572 | return -EPERM; | |
573 | if (btrfs_check_sticky(dir, victim->d_inode)|| | |
574 | IS_APPEND(victim->d_inode)|| | |
575 | IS_IMMUTABLE(victim->d_inode) || IS_SWAPFILE(victim->d_inode)) | |
576 | return -EPERM; | |
577 | if (isdir) { | |
578 | if (!S_ISDIR(victim->d_inode->i_mode)) | |
579 | return -ENOTDIR; | |
580 | if (IS_ROOT(victim)) | |
581 | return -EBUSY; | |
582 | } else if (S_ISDIR(victim->d_inode->i_mode)) | |
583 | return -EISDIR; | |
584 | if (IS_DEADDIR(dir)) | |
585 | return -ENOENT; | |
586 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
587 | return -EBUSY; | |
588 | return 0; | |
589 | } | |
590 | ||
cb8e7090 CH |
591 | /* copy of may_create in fs/namei.c() */ |
592 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
593 | { | |
594 | if (child->d_inode) | |
595 | return -EEXIST; | |
596 | if (IS_DEADDIR(dir)) | |
597 | return -ENOENT; | |
598 | return inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
599 | } | |
600 | ||
601 | /* | |
602 | * Create a new subvolume below @parent. This is largely modeled after | |
603 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
604 | * inside this filesystem so it's quite a bit simpler. | |
605 | */ | |
76dda93c YZ |
606 | static noinline int btrfs_mksubvol(struct path *parent, |
607 | char *name, int namelen, | |
72fd032e | 608 | struct btrfs_root *snap_src, |
b83cc969 | 609 | u64 *async_transid, bool readonly) |
cb8e7090 | 610 | { |
76dda93c | 611 | struct inode *dir = parent->dentry->d_inode; |
cb8e7090 CH |
612 | struct dentry *dentry; |
613 | int error; | |
614 | ||
76dda93c | 615 | mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); |
cb8e7090 CH |
616 | |
617 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
618 | error = PTR_ERR(dentry); | |
619 | if (IS_ERR(dentry)) | |
620 | goto out_unlock; | |
621 | ||
622 | error = -EEXIST; | |
623 | if (dentry->d_inode) | |
624 | goto out_dput; | |
625 | ||
cb8e7090 CH |
626 | error = mnt_want_write(parent->mnt); |
627 | if (error) | |
628 | goto out_dput; | |
629 | ||
76dda93c | 630 | error = btrfs_may_create(dir, dentry); |
cb8e7090 CH |
631 | if (error) |
632 | goto out_drop_write; | |
633 | ||
76dda93c YZ |
634 | down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem); |
635 | ||
636 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
637 | goto out_up_read; | |
638 | ||
3de4586c | 639 | if (snap_src) { |
72fd032e | 640 | error = create_snapshot(snap_src, dentry, |
b83cc969 | 641 | name, namelen, async_transid, readonly); |
3de4586c | 642 | } else { |
76dda93c | 643 | error = create_subvol(BTRFS_I(dir)->root, dentry, |
72fd032e | 644 | name, namelen, async_transid); |
3de4586c | 645 | } |
76dda93c YZ |
646 | if (!error) |
647 | fsnotify_mkdir(dir, dentry); | |
648 | out_up_read: | |
649 | up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem); | |
cb8e7090 CH |
650 | out_drop_write: |
651 | mnt_drop_write(parent->mnt); | |
652 | out_dput: | |
653 | dput(dentry); | |
654 | out_unlock: | |
76dda93c | 655 | mutex_unlock(&dir->i_mutex); |
cb8e7090 CH |
656 | return error; |
657 | } | |
658 | ||
940100a4 | 659 | static int should_defrag_range(struct inode *inode, u64 start, u64 len, |
1e701a32 CM |
660 | int thresh, u64 *last_len, u64 *skip, |
661 | u64 *defrag_end) | |
940100a4 CM |
662 | { |
663 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
664 | struct extent_map *em = NULL; | |
665 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
666 | int ret = 1; | |
667 | ||
1e701a32 CM |
668 | |
669 | if (thresh == 0) | |
670 | thresh = 256 * 1024; | |
671 | ||
940100a4 CM |
672 | /* |
673 | * make sure that once we start defragging and extent, we keep on | |
674 | * defragging it | |
675 | */ | |
676 | if (start < *defrag_end) | |
677 | return 1; | |
678 | ||
679 | *skip = 0; | |
680 | ||
681 | /* | |
682 | * hopefully we have this extent in the tree already, try without | |
683 | * the full extent lock | |
684 | */ | |
685 | read_lock(&em_tree->lock); | |
686 | em = lookup_extent_mapping(em_tree, start, len); | |
687 | read_unlock(&em_tree->lock); | |
688 | ||
689 | if (!em) { | |
690 | /* get the big lock and read metadata off disk */ | |
691 | lock_extent(io_tree, start, start + len - 1, GFP_NOFS); | |
692 | em = btrfs_get_extent(inode, NULL, 0, start, len, 0); | |
693 | unlock_extent(io_tree, start, start + len - 1, GFP_NOFS); | |
694 | ||
6cf8bfbf | 695 | if (IS_ERR(em)) |
940100a4 CM |
696 | return 0; |
697 | } | |
698 | ||
699 | /* this will cover holes, and inline extents */ | |
700 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) | |
701 | ret = 0; | |
702 | ||
703 | /* | |
704 | * we hit a real extent, if it is big don't bother defragging it again | |
705 | */ | |
1e701a32 | 706 | if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh) |
940100a4 CM |
707 | ret = 0; |
708 | ||
709 | /* | |
710 | * last_len ends up being a counter of how many bytes we've defragged. | |
711 | * every time we choose not to defrag an extent, we reset *last_len | |
712 | * so that the next tiny extent will force a defrag. | |
713 | * | |
714 | * The end result of this is that tiny extents before a single big | |
715 | * extent will force at least part of that big extent to be defragged. | |
716 | */ | |
717 | if (ret) { | |
718 | *last_len += len; | |
719 | *defrag_end = extent_map_end(em); | |
720 | } else { | |
721 | *last_len = 0; | |
722 | *skip = extent_map_end(em); | |
723 | *defrag_end = 0; | |
724 | } | |
725 | ||
726 | free_extent_map(em); | |
727 | return ret; | |
728 | } | |
729 | ||
1e701a32 CM |
730 | static int btrfs_defrag_file(struct file *file, |
731 | struct btrfs_ioctl_defrag_range_args *range) | |
f46b5a66 CH |
732 | { |
733 | struct inode *inode = fdentry(file)->d_inode; | |
734 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
735 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
3eaa2885 | 736 | struct btrfs_ordered_extent *ordered; |
f46b5a66 | 737 | struct page *page; |
1a419d85 | 738 | struct btrfs_super_block *disk_super; |
f46b5a66 CH |
739 | unsigned long last_index; |
740 | unsigned long ra_pages = root->fs_info->bdi.ra_pages; | |
741 | unsigned long total_read = 0; | |
1a419d85 | 742 | u64 features; |
f46b5a66 CH |
743 | u64 page_start; |
744 | u64 page_end; | |
940100a4 CM |
745 | u64 last_len = 0; |
746 | u64 skip = 0; | |
747 | u64 defrag_end = 0; | |
f46b5a66 CH |
748 | unsigned long i; |
749 | int ret; | |
1a419d85 LZ |
750 | int compress_type = BTRFS_COMPRESS_ZLIB; |
751 | ||
752 | if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) { | |
753 | if (range->compress_type > BTRFS_COMPRESS_TYPES) | |
754 | return -EINVAL; | |
755 | if (range->compress_type) | |
756 | compress_type = range->compress_type; | |
757 | } | |
f46b5a66 | 758 | |
940100a4 CM |
759 | if (inode->i_size == 0) |
760 | return 0; | |
761 | ||
1e701a32 CM |
762 | if (range->start + range->len > range->start) { |
763 | last_index = min_t(u64, inode->i_size - 1, | |
764 | range->start + range->len - 1) >> PAGE_CACHE_SHIFT; | |
765 | } else { | |
766 | last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT; | |
767 | } | |
768 | ||
769 | i = range->start >> PAGE_CACHE_SHIFT; | |
940100a4 CM |
770 | while (i <= last_index) { |
771 | if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, | |
1e701a32 CM |
772 | PAGE_CACHE_SIZE, |
773 | range->extent_thresh, | |
774 | &last_len, &skip, | |
940100a4 CM |
775 | &defrag_end)) { |
776 | unsigned long next; | |
777 | /* | |
778 | * the should_defrag function tells us how much to skip | |
779 | * bump our counter by the suggested amount | |
780 | */ | |
781 | next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; | |
782 | i = max(i + 1, next); | |
783 | continue; | |
784 | } | |
f46b5a66 | 785 | |
f46b5a66 CH |
786 | if (total_read % ra_pages == 0) { |
787 | btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i, | |
788 | min(last_index, i + ra_pages - 1)); | |
789 | } | |
790 | total_read++; | |
940100a4 | 791 | mutex_lock(&inode->i_mutex); |
1e701a32 | 792 | if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) |
1a419d85 | 793 | BTRFS_I(inode)->force_compress = compress_type; |
940100a4 | 794 | |
0ca1f7ce YZ |
795 | ret = btrfs_delalloc_reserve_space(inode, PAGE_CACHE_SIZE); |
796 | if (ret) | |
797 | goto err_unlock; | |
3eaa2885 | 798 | again: |
940100a4 CM |
799 | if (inode->i_size == 0 || |
800 | i > ((inode->i_size - 1) >> PAGE_CACHE_SHIFT)) { | |
801 | ret = 0; | |
802 | goto err_reservations; | |
803 | } | |
804 | ||
f46b5a66 | 805 | page = grab_cache_page(inode->i_mapping, i); |
0ca1f7ce YZ |
806 | if (!page) { |
807 | ret = -ENOMEM; | |
940100a4 | 808 | goto err_reservations; |
0ca1f7ce | 809 | } |
940100a4 | 810 | |
f46b5a66 CH |
811 | if (!PageUptodate(page)) { |
812 | btrfs_readpage(NULL, page); | |
813 | lock_page(page); | |
814 | if (!PageUptodate(page)) { | |
815 | unlock_page(page); | |
816 | page_cache_release(page); | |
0ca1f7ce | 817 | ret = -EIO; |
940100a4 | 818 | goto err_reservations; |
f46b5a66 CH |
819 | } |
820 | } | |
821 | ||
940100a4 CM |
822 | if (page->mapping != inode->i_mapping) { |
823 | unlock_page(page); | |
824 | page_cache_release(page); | |
825 | goto again; | |
826 | } | |
827 | ||
f46b5a66 | 828 | wait_on_page_writeback(page); |
f46b5a66 | 829 | |
940100a4 | 830 | if (PageDirty(page)) { |
0ca1f7ce | 831 | btrfs_delalloc_release_space(inode, PAGE_CACHE_SIZE); |
940100a4 CM |
832 | goto loop_unlock; |
833 | } | |
834 | ||
f46b5a66 CH |
835 | page_start = (u64)page->index << PAGE_CACHE_SHIFT; |
836 | page_end = page_start + PAGE_CACHE_SIZE - 1; | |
f46b5a66 | 837 | lock_extent(io_tree, page_start, page_end, GFP_NOFS); |
3eaa2885 CM |
838 | |
839 | ordered = btrfs_lookup_ordered_extent(inode, page_start); | |
840 | if (ordered) { | |
841 | unlock_extent(io_tree, page_start, page_end, GFP_NOFS); | |
842 | unlock_page(page); | |
843 | page_cache_release(page); | |
844 | btrfs_start_ordered_extent(inode, ordered, 1); | |
845 | btrfs_put_ordered_extent(ordered); | |
846 | goto again; | |
847 | } | |
848 | set_page_extent_mapped(page); | |
849 | ||
f87f057b CM |
850 | /* |
851 | * this makes sure page_mkwrite is called on the | |
852 | * page if it is dirtied again later | |
853 | */ | |
854 | clear_page_dirty_for_io(page); | |
940100a4 CM |
855 | clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, |
856 | page_end, EXTENT_DIRTY | EXTENT_DELALLOC | | |
857 | EXTENT_DO_ACCOUNTING, GFP_NOFS); | |
f87f057b | 858 | |
2ac55d41 | 859 | btrfs_set_extent_delalloc(inode, page_start, page_end, NULL); |
940100a4 | 860 | ClearPageChecked(page); |
f46b5a66 | 861 | set_page_dirty(page); |
a1ed835e | 862 | unlock_extent(io_tree, page_start, page_end, GFP_NOFS); |
940100a4 CM |
863 | |
864 | loop_unlock: | |
f46b5a66 CH |
865 | unlock_page(page); |
866 | page_cache_release(page); | |
940100a4 CM |
867 | mutex_unlock(&inode->i_mutex); |
868 | ||
f46b5a66 | 869 | balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1); |
940100a4 | 870 | i++; |
f46b5a66 CH |
871 | } |
872 | ||
1e701a32 CM |
873 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) |
874 | filemap_flush(inode->i_mapping); | |
875 | ||
876 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
877 | /* the filemap_flush will queue IO into the worker threads, but | |
878 | * we have to make sure the IO is actually started and that | |
879 | * ordered extents get created before we return | |
880 | */ | |
881 | atomic_inc(&root->fs_info->async_submit_draining); | |
882 | while (atomic_read(&root->fs_info->nr_async_submits) || | |
883 | atomic_read(&root->fs_info->async_delalloc_pages)) { | |
884 | wait_event(root->fs_info->async_submit_wait, | |
885 | (atomic_read(&root->fs_info->nr_async_submits) == 0 && | |
886 | atomic_read(&root->fs_info->async_delalloc_pages) == 0)); | |
887 | } | |
888 | atomic_dec(&root->fs_info->async_submit_draining); | |
889 | ||
890 | mutex_lock(&inode->i_mutex); | |
261507a0 | 891 | BTRFS_I(inode)->force_compress = BTRFS_COMPRESS_NONE; |
1e701a32 CM |
892 | mutex_unlock(&inode->i_mutex); |
893 | } | |
894 | ||
1a419d85 LZ |
895 | disk_super = &root->fs_info->super_copy; |
896 | features = btrfs_super_incompat_flags(disk_super); | |
897 | if (range->compress_type == BTRFS_COMPRESS_LZO) { | |
898 | features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO; | |
899 | btrfs_set_super_incompat_flags(disk_super, features); | |
900 | } | |
901 | ||
f46b5a66 | 902 | return 0; |
940100a4 CM |
903 | |
904 | err_reservations: | |
0ca1f7ce YZ |
905 | btrfs_delalloc_release_space(inode, PAGE_CACHE_SIZE); |
906 | err_unlock: | |
940100a4 | 907 | mutex_unlock(&inode->i_mutex); |
940100a4 | 908 | return ret; |
f46b5a66 CH |
909 | } |
910 | ||
76dda93c YZ |
911 | static noinline int btrfs_ioctl_resize(struct btrfs_root *root, |
912 | void __user *arg) | |
f46b5a66 CH |
913 | { |
914 | u64 new_size; | |
915 | u64 old_size; | |
916 | u64 devid = 1; | |
917 | struct btrfs_ioctl_vol_args *vol_args; | |
918 | struct btrfs_trans_handle *trans; | |
919 | struct btrfs_device *device = NULL; | |
920 | char *sizestr; | |
921 | char *devstr = NULL; | |
922 | int ret = 0; | |
f46b5a66 CH |
923 | int mod = 0; |
924 | ||
c146afad YZ |
925 | if (root->fs_info->sb->s_flags & MS_RDONLY) |
926 | return -EROFS; | |
927 | ||
e441d54d CM |
928 | if (!capable(CAP_SYS_ADMIN)) |
929 | return -EPERM; | |
930 | ||
dae7b665 LZ |
931 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
932 | if (IS_ERR(vol_args)) | |
933 | return PTR_ERR(vol_args); | |
5516e595 MF |
934 | |
935 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 936 | |
7d9eb12c | 937 | mutex_lock(&root->fs_info->volume_mutex); |
f46b5a66 CH |
938 | sizestr = vol_args->name; |
939 | devstr = strchr(sizestr, ':'); | |
940 | if (devstr) { | |
941 | char *end; | |
942 | sizestr = devstr + 1; | |
943 | *devstr = '\0'; | |
944 | devstr = vol_args->name; | |
945 | devid = simple_strtoull(devstr, &end, 10); | |
21380931 JB |
946 | printk(KERN_INFO "resizing devid %llu\n", |
947 | (unsigned long long)devid); | |
f46b5a66 | 948 | } |
2b82032c | 949 | device = btrfs_find_device(root, devid, NULL, NULL); |
f46b5a66 | 950 | if (!device) { |
21380931 JB |
951 | printk(KERN_INFO "resizer unable to find device %llu\n", |
952 | (unsigned long long)devid); | |
f46b5a66 CH |
953 | ret = -EINVAL; |
954 | goto out_unlock; | |
955 | } | |
956 | if (!strcmp(sizestr, "max")) | |
957 | new_size = device->bdev->bd_inode->i_size; | |
958 | else { | |
959 | if (sizestr[0] == '-') { | |
960 | mod = -1; | |
961 | sizestr++; | |
962 | } else if (sizestr[0] == '+') { | |
963 | mod = 1; | |
964 | sizestr++; | |
965 | } | |
91748467 | 966 | new_size = memparse(sizestr, NULL); |
f46b5a66 CH |
967 | if (new_size == 0) { |
968 | ret = -EINVAL; | |
969 | goto out_unlock; | |
970 | } | |
971 | } | |
972 | ||
973 | old_size = device->total_bytes; | |
974 | ||
975 | if (mod < 0) { | |
976 | if (new_size > old_size) { | |
977 | ret = -EINVAL; | |
978 | goto out_unlock; | |
979 | } | |
980 | new_size = old_size - new_size; | |
981 | } else if (mod > 0) { | |
982 | new_size = old_size + new_size; | |
983 | } | |
984 | ||
985 | if (new_size < 256 * 1024 * 1024) { | |
986 | ret = -EINVAL; | |
987 | goto out_unlock; | |
988 | } | |
989 | if (new_size > device->bdev->bd_inode->i_size) { | |
990 | ret = -EFBIG; | |
991 | goto out_unlock; | |
992 | } | |
993 | ||
994 | do_div(new_size, root->sectorsize); | |
995 | new_size *= root->sectorsize; | |
996 | ||
997 | printk(KERN_INFO "new size for %s is %llu\n", | |
998 | device->name, (unsigned long long)new_size); | |
999 | ||
1000 | if (new_size > old_size) { | |
a22285a6 | 1001 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1002 | if (IS_ERR(trans)) { |
1003 | ret = PTR_ERR(trans); | |
1004 | goto out_unlock; | |
1005 | } | |
f46b5a66 CH |
1006 | ret = btrfs_grow_device(trans, device, new_size); |
1007 | btrfs_commit_transaction(trans, root); | |
1008 | } else { | |
1009 | ret = btrfs_shrink_device(device, new_size); | |
1010 | } | |
1011 | ||
1012 | out_unlock: | |
7d9eb12c | 1013 | mutex_unlock(&root->fs_info->volume_mutex); |
f46b5a66 CH |
1014 | kfree(vol_args); |
1015 | return ret; | |
1016 | } | |
1017 | ||
72fd032e SW |
1018 | static noinline int btrfs_ioctl_snap_create_transid(struct file *file, |
1019 | char *name, | |
1020 | unsigned long fd, | |
1021 | int subvol, | |
b83cc969 LZ |
1022 | u64 *transid, |
1023 | bool readonly) | |
f46b5a66 | 1024 | { |
cb8e7090 | 1025 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; |
3de4586c | 1026 | struct file *src_file; |
f46b5a66 | 1027 | int namelen; |
3de4586c | 1028 | int ret = 0; |
f46b5a66 | 1029 | |
c146afad YZ |
1030 | if (root->fs_info->sb->s_flags & MS_RDONLY) |
1031 | return -EROFS; | |
1032 | ||
72fd032e SW |
1033 | namelen = strlen(name); |
1034 | if (strchr(name, '/')) { | |
f46b5a66 CH |
1035 | ret = -EINVAL; |
1036 | goto out; | |
1037 | } | |
1038 | ||
3de4586c | 1039 | if (subvol) { |
72fd032e | 1040 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
b83cc969 | 1041 | NULL, transid, readonly); |
cb8e7090 | 1042 | } else { |
3de4586c | 1043 | struct inode *src_inode; |
72fd032e | 1044 | src_file = fget(fd); |
3de4586c CM |
1045 | if (!src_file) { |
1046 | ret = -EINVAL; | |
1047 | goto out; | |
1048 | } | |
1049 | ||
1050 | src_inode = src_file->f_path.dentry->d_inode; | |
1051 | if (src_inode->i_sb != file->f_path.dentry->d_inode->i_sb) { | |
d397712b CM |
1052 | printk(KERN_INFO "btrfs: Snapshot src from " |
1053 | "another FS\n"); | |
3de4586c CM |
1054 | ret = -EINVAL; |
1055 | fput(src_file); | |
1056 | goto out; | |
1057 | } | |
72fd032e SW |
1058 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
1059 | BTRFS_I(src_inode)->root, | |
b83cc969 | 1060 | transid, readonly); |
3de4586c | 1061 | fput(src_file); |
cb8e7090 | 1062 | } |
f46b5a66 | 1063 | out: |
72fd032e SW |
1064 | return ret; |
1065 | } | |
1066 | ||
1067 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1068 | void __user *arg, int subvol) |
72fd032e | 1069 | { |
fa0d2b9b | 1070 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1071 | int ret; |
1072 | ||
fa0d2b9b LZ |
1073 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1074 | if (IS_ERR(vol_args)) | |
1075 | return PTR_ERR(vol_args); | |
1076 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1077 | |
fa0d2b9b | 1078 | ret = btrfs_ioctl_snap_create_transid(file, vol_args->name, |
b83cc969 LZ |
1079 | vol_args->fd, subvol, |
1080 | NULL, false); | |
fdfb1e4f | 1081 | |
fa0d2b9b LZ |
1082 | kfree(vol_args); |
1083 | return ret; | |
1084 | } | |
fdfb1e4f | 1085 | |
fa0d2b9b LZ |
1086 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1087 | void __user *arg, int subvol) | |
1088 | { | |
1089 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1090 | int ret; | |
1091 | u64 transid = 0; | |
1092 | u64 *ptr = NULL; | |
b83cc969 | 1093 | bool readonly = false; |
75eaa0e2 | 1094 | |
fa0d2b9b LZ |
1095 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1096 | if (IS_ERR(vol_args)) | |
1097 | return PTR_ERR(vol_args); | |
1098 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1099 | |
b83cc969 LZ |
1100 | if (vol_args->flags & |
1101 | ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY)) { | |
1102 | ret = -EOPNOTSUPP; | |
fa0d2b9b | 1103 | goto out; |
72fd032e | 1104 | } |
fa0d2b9b LZ |
1105 | |
1106 | if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC) | |
1107 | ptr = &transid; | |
b83cc969 LZ |
1108 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1109 | readonly = true; | |
fa0d2b9b LZ |
1110 | |
1111 | ret = btrfs_ioctl_snap_create_transid(file, vol_args->name, | |
b83cc969 LZ |
1112 | vol_args->fd, subvol, |
1113 | ptr, readonly); | |
fa0d2b9b LZ |
1114 | |
1115 | if (ret == 0 && ptr && | |
1116 | copy_to_user(arg + | |
1117 | offsetof(struct btrfs_ioctl_vol_args_v2, | |
1118 | transid), ptr, sizeof(*ptr))) | |
1119 | ret = -EFAULT; | |
fdfb1e4f | 1120 | out: |
f46b5a66 CH |
1121 | kfree(vol_args); |
1122 | return ret; | |
1123 | } | |
1124 | ||
0caa102d LZ |
1125 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1126 | void __user *arg) | |
1127 | { | |
1128 | struct inode *inode = fdentry(file)->d_inode; | |
1129 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
1130 | int ret = 0; | |
1131 | u64 flags = 0; | |
1132 | ||
33345d01 | 1133 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1134 | return -EINVAL; |
1135 | ||
1136 | down_read(&root->fs_info->subvol_sem); | |
1137 | if (btrfs_root_readonly(root)) | |
1138 | flags |= BTRFS_SUBVOL_RDONLY; | |
1139 | up_read(&root->fs_info->subvol_sem); | |
1140 | ||
1141 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1142 | ret = -EFAULT; | |
1143 | ||
1144 | return ret; | |
1145 | } | |
1146 | ||
1147 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1148 | void __user *arg) | |
1149 | { | |
1150 | struct inode *inode = fdentry(file)->d_inode; | |
1151 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
1152 | struct btrfs_trans_handle *trans; | |
1153 | u64 root_flags; | |
1154 | u64 flags; | |
1155 | int ret = 0; | |
1156 | ||
1157 | if (root->fs_info->sb->s_flags & MS_RDONLY) | |
1158 | return -EROFS; | |
1159 | ||
33345d01 | 1160 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1161 | return -EINVAL; |
1162 | ||
1163 | if (copy_from_user(&flags, arg, sizeof(flags))) | |
1164 | return -EFAULT; | |
1165 | ||
b4dc2b8c | 1166 | if (flags & BTRFS_SUBVOL_CREATE_ASYNC) |
0caa102d LZ |
1167 | return -EINVAL; |
1168 | ||
1169 | if (flags & ~BTRFS_SUBVOL_RDONLY) | |
1170 | return -EOPNOTSUPP; | |
1171 | ||
2e149670 | 1172 | if (!inode_owner_or_capable(inode)) |
b4dc2b8c LZ |
1173 | return -EACCES; |
1174 | ||
0caa102d LZ |
1175 | down_write(&root->fs_info->subvol_sem); |
1176 | ||
1177 | /* nothing to do */ | |
1178 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
1179 | goto out; | |
1180 | ||
1181 | root_flags = btrfs_root_flags(&root->root_item); | |
1182 | if (flags & BTRFS_SUBVOL_RDONLY) | |
1183 | btrfs_set_root_flags(&root->root_item, | |
1184 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
1185 | else | |
1186 | btrfs_set_root_flags(&root->root_item, | |
1187 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); | |
1188 | ||
1189 | trans = btrfs_start_transaction(root, 1); | |
1190 | if (IS_ERR(trans)) { | |
1191 | ret = PTR_ERR(trans); | |
1192 | goto out_reset; | |
1193 | } | |
1194 | ||
b4dc2b8c | 1195 | ret = btrfs_update_root(trans, root->fs_info->tree_root, |
0caa102d LZ |
1196 | &root->root_key, &root->root_item); |
1197 | ||
1198 | btrfs_commit_transaction(trans, root); | |
1199 | out_reset: | |
1200 | if (ret) | |
1201 | btrfs_set_root_flags(&root->root_item, root_flags); | |
1202 | out: | |
1203 | up_write(&root->fs_info->subvol_sem); | |
1204 | return ret; | |
1205 | } | |
1206 | ||
76dda93c YZ |
1207 | /* |
1208 | * helper to check if the subvolume references other subvolumes | |
1209 | */ | |
1210 | static noinline int may_destroy_subvol(struct btrfs_root *root) | |
1211 | { | |
1212 | struct btrfs_path *path; | |
1213 | struct btrfs_key key; | |
1214 | int ret; | |
1215 | ||
1216 | path = btrfs_alloc_path(); | |
1217 | if (!path) | |
1218 | return -ENOMEM; | |
1219 | ||
1220 | key.objectid = root->root_key.objectid; | |
1221 | key.type = BTRFS_ROOT_REF_KEY; | |
1222 | key.offset = (u64)-1; | |
1223 | ||
1224 | ret = btrfs_search_slot(NULL, root->fs_info->tree_root, | |
1225 | &key, path, 0, 0); | |
1226 | if (ret < 0) | |
1227 | goto out; | |
1228 | BUG_ON(ret == 0); | |
1229 | ||
1230 | ret = 0; | |
1231 | if (path->slots[0] > 0) { | |
1232 | path->slots[0]--; | |
1233 | btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]); | |
1234 | if (key.objectid == root->root_key.objectid && | |
1235 | key.type == BTRFS_ROOT_REF_KEY) | |
1236 | ret = -ENOTEMPTY; | |
1237 | } | |
1238 | out: | |
1239 | btrfs_free_path(path); | |
1240 | return ret; | |
1241 | } | |
1242 | ||
ac8e9819 CM |
1243 | static noinline int key_in_sk(struct btrfs_key *key, |
1244 | struct btrfs_ioctl_search_key *sk) | |
1245 | { | |
abc6e134 CM |
1246 | struct btrfs_key test; |
1247 | int ret; | |
1248 | ||
1249 | test.objectid = sk->min_objectid; | |
1250 | test.type = sk->min_type; | |
1251 | test.offset = sk->min_offset; | |
1252 | ||
1253 | ret = btrfs_comp_cpu_keys(key, &test); | |
1254 | if (ret < 0) | |
ac8e9819 | 1255 | return 0; |
abc6e134 CM |
1256 | |
1257 | test.objectid = sk->max_objectid; | |
1258 | test.type = sk->max_type; | |
1259 | test.offset = sk->max_offset; | |
1260 | ||
1261 | ret = btrfs_comp_cpu_keys(key, &test); | |
1262 | if (ret > 0) | |
ac8e9819 CM |
1263 | return 0; |
1264 | return 1; | |
1265 | } | |
1266 | ||
1267 | static noinline int copy_to_sk(struct btrfs_root *root, | |
1268 | struct btrfs_path *path, | |
1269 | struct btrfs_key *key, | |
1270 | struct btrfs_ioctl_search_key *sk, | |
1271 | char *buf, | |
1272 | unsigned long *sk_offset, | |
1273 | int *num_found) | |
1274 | { | |
1275 | u64 found_transid; | |
1276 | struct extent_buffer *leaf; | |
1277 | struct btrfs_ioctl_search_header sh; | |
1278 | unsigned long item_off; | |
1279 | unsigned long item_len; | |
1280 | int nritems; | |
1281 | int i; | |
1282 | int slot; | |
ac8e9819 CM |
1283 | int ret = 0; |
1284 | ||
1285 | leaf = path->nodes[0]; | |
1286 | slot = path->slots[0]; | |
1287 | nritems = btrfs_header_nritems(leaf); | |
1288 | ||
1289 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
1290 | i = nritems; | |
1291 | goto advance_key; | |
1292 | } | |
1293 | found_transid = btrfs_header_generation(leaf); | |
1294 | ||
1295 | for (i = slot; i < nritems; i++) { | |
1296 | item_off = btrfs_item_ptr_offset(leaf, i); | |
1297 | item_len = btrfs_item_size_nr(leaf, i); | |
1298 | ||
1299 | if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE) | |
1300 | item_len = 0; | |
1301 | ||
1302 | if (sizeof(sh) + item_len + *sk_offset > | |
1303 | BTRFS_SEARCH_ARGS_BUFSIZE) { | |
1304 | ret = 1; | |
1305 | goto overflow; | |
1306 | } | |
1307 | ||
1308 | btrfs_item_key_to_cpu(leaf, key, i); | |
1309 | if (!key_in_sk(key, sk)) | |
1310 | continue; | |
1311 | ||
1312 | sh.objectid = key->objectid; | |
1313 | sh.offset = key->offset; | |
1314 | sh.type = key->type; | |
1315 | sh.len = item_len; | |
1316 | sh.transid = found_transid; | |
1317 | ||
1318 | /* copy search result header */ | |
1319 | memcpy(buf + *sk_offset, &sh, sizeof(sh)); | |
1320 | *sk_offset += sizeof(sh); | |
1321 | ||
1322 | if (item_len) { | |
1323 | char *p = buf + *sk_offset; | |
1324 | /* copy the item */ | |
1325 | read_extent_buffer(leaf, p, | |
1326 | item_off, item_len); | |
1327 | *sk_offset += item_len; | |
ac8e9819 | 1328 | } |
e2156867 | 1329 | (*num_found)++; |
ac8e9819 CM |
1330 | |
1331 | if (*num_found >= sk->nr_items) | |
1332 | break; | |
1333 | } | |
1334 | advance_key: | |
abc6e134 CM |
1335 | ret = 0; |
1336 | if (key->offset < (u64)-1 && key->offset < sk->max_offset) | |
ac8e9819 | 1337 | key->offset++; |
abc6e134 CM |
1338 | else if (key->type < (u8)-1 && key->type < sk->max_type) { |
1339 | key->offset = 0; | |
ac8e9819 | 1340 | key->type++; |
abc6e134 CM |
1341 | } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) { |
1342 | key->offset = 0; | |
1343 | key->type = 0; | |
ac8e9819 | 1344 | key->objectid++; |
abc6e134 CM |
1345 | } else |
1346 | ret = 1; | |
ac8e9819 | 1347 | overflow: |
ac8e9819 CM |
1348 | return ret; |
1349 | } | |
1350 | ||
1351 | static noinline int search_ioctl(struct inode *inode, | |
1352 | struct btrfs_ioctl_search_args *args) | |
1353 | { | |
1354 | struct btrfs_root *root; | |
1355 | struct btrfs_key key; | |
1356 | struct btrfs_key max_key; | |
1357 | struct btrfs_path *path; | |
1358 | struct btrfs_ioctl_search_key *sk = &args->key; | |
1359 | struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info; | |
1360 | int ret; | |
1361 | int num_found = 0; | |
1362 | unsigned long sk_offset = 0; | |
1363 | ||
1364 | path = btrfs_alloc_path(); | |
1365 | if (!path) | |
1366 | return -ENOMEM; | |
1367 | ||
1368 | if (sk->tree_id == 0) { | |
1369 | /* search the root of the inode that was passed */ | |
1370 | root = BTRFS_I(inode)->root; | |
1371 | } else { | |
1372 | key.objectid = sk->tree_id; | |
1373 | key.type = BTRFS_ROOT_ITEM_KEY; | |
1374 | key.offset = (u64)-1; | |
1375 | root = btrfs_read_fs_root_no_name(info, &key); | |
1376 | if (IS_ERR(root)) { | |
1377 | printk(KERN_ERR "could not find root %llu\n", | |
1378 | sk->tree_id); | |
1379 | btrfs_free_path(path); | |
1380 | return -ENOENT; | |
1381 | } | |
1382 | } | |
1383 | ||
1384 | key.objectid = sk->min_objectid; | |
1385 | key.type = sk->min_type; | |
1386 | key.offset = sk->min_offset; | |
1387 | ||
1388 | max_key.objectid = sk->max_objectid; | |
1389 | max_key.type = sk->max_type; | |
1390 | max_key.offset = sk->max_offset; | |
1391 | ||
1392 | path->keep_locks = 1; | |
1393 | ||
1394 | while(1) { | |
1395 | ret = btrfs_search_forward(root, &key, &max_key, path, 0, | |
1396 | sk->min_transid); | |
1397 | if (ret != 0) { | |
1398 | if (ret > 0) | |
1399 | ret = 0; | |
1400 | goto err; | |
1401 | } | |
1402 | ret = copy_to_sk(root, path, &key, sk, args->buf, | |
1403 | &sk_offset, &num_found); | |
b3b4aa74 | 1404 | btrfs_release_path(path); |
ac8e9819 CM |
1405 | if (ret || num_found >= sk->nr_items) |
1406 | break; | |
1407 | ||
1408 | } | |
1409 | ret = 0; | |
1410 | err: | |
1411 | sk->nr_items = num_found; | |
1412 | btrfs_free_path(path); | |
1413 | return ret; | |
1414 | } | |
1415 | ||
1416 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
1417 | void __user *argp) | |
1418 | { | |
1419 | struct btrfs_ioctl_search_args *args; | |
1420 | struct inode *inode; | |
1421 | int ret; | |
1422 | ||
1423 | if (!capable(CAP_SYS_ADMIN)) | |
1424 | return -EPERM; | |
1425 | ||
2354d08f JL |
1426 | args = memdup_user(argp, sizeof(*args)); |
1427 | if (IS_ERR(args)) | |
1428 | return PTR_ERR(args); | |
ac8e9819 | 1429 | |
ac8e9819 CM |
1430 | inode = fdentry(file)->d_inode; |
1431 | ret = search_ioctl(inode, args); | |
1432 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
1433 | ret = -EFAULT; | |
1434 | kfree(args); | |
1435 | return ret; | |
1436 | } | |
1437 | ||
98d377a0 | 1438 | /* |
ac8e9819 CM |
1439 | * Search INODE_REFs to identify path name of 'dirid' directory |
1440 | * in a 'tree_id' tree. and sets path name to 'name'. | |
1441 | */ | |
98d377a0 TH |
1442 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
1443 | u64 tree_id, u64 dirid, char *name) | |
1444 | { | |
1445 | struct btrfs_root *root; | |
1446 | struct btrfs_key key; | |
ac8e9819 | 1447 | char *ptr; |
98d377a0 TH |
1448 | int ret = -1; |
1449 | int slot; | |
1450 | int len; | |
1451 | int total_len = 0; | |
1452 | struct btrfs_inode_ref *iref; | |
1453 | struct extent_buffer *l; | |
1454 | struct btrfs_path *path; | |
1455 | ||
1456 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
1457 | name[0]='\0'; | |
1458 | return 0; | |
1459 | } | |
1460 | ||
1461 | path = btrfs_alloc_path(); | |
1462 | if (!path) | |
1463 | return -ENOMEM; | |
1464 | ||
ac8e9819 | 1465 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX]; |
98d377a0 TH |
1466 | |
1467 | key.objectid = tree_id; | |
1468 | key.type = BTRFS_ROOT_ITEM_KEY; | |
1469 | key.offset = (u64)-1; | |
1470 | root = btrfs_read_fs_root_no_name(info, &key); | |
1471 | if (IS_ERR(root)) { | |
1472 | printk(KERN_ERR "could not find root %llu\n", tree_id); | |
8ad6fcab CM |
1473 | ret = -ENOENT; |
1474 | goto out; | |
98d377a0 TH |
1475 | } |
1476 | ||
1477 | key.objectid = dirid; | |
1478 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 1479 | key.offset = (u64)-1; |
98d377a0 TH |
1480 | |
1481 | while(1) { | |
1482 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
1483 | if (ret < 0) | |
1484 | goto out; | |
1485 | ||
1486 | l = path->nodes[0]; | |
1487 | slot = path->slots[0]; | |
8ad6fcab CM |
1488 | if (ret > 0 && slot > 0) |
1489 | slot--; | |
98d377a0 TH |
1490 | btrfs_item_key_to_cpu(l, &key, slot); |
1491 | ||
1492 | if (ret > 0 && (key.objectid != dirid || | |
ac8e9819 CM |
1493 | key.type != BTRFS_INODE_REF_KEY)) { |
1494 | ret = -ENOENT; | |
98d377a0 | 1495 | goto out; |
ac8e9819 | 1496 | } |
98d377a0 TH |
1497 | |
1498 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); | |
1499 | len = btrfs_inode_ref_name_len(l, iref); | |
1500 | ptr -= len + 1; | |
1501 | total_len += len + 1; | |
ac8e9819 | 1502 | if (ptr < name) |
98d377a0 TH |
1503 | goto out; |
1504 | ||
1505 | *(ptr + len) = '/'; | |
1506 | read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len); | |
1507 | ||
1508 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
1509 | break; | |
1510 | ||
b3b4aa74 | 1511 | btrfs_release_path(path); |
98d377a0 | 1512 | key.objectid = key.offset; |
8ad6fcab | 1513 | key.offset = (u64)-1; |
98d377a0 TH |
1514 | dirid = key.objectid; |
1515 | ||
1516 | } | |
ac8e9819 | 1517 | if (ptr < name) |
98d377a0 | 1518 | goto out; |
ac8e9819 | 1519 | memcpy(name, ptr, total_len); |
98d377a0 TH |
1520 | name[total_len]='\0'; |
1521 | ret = 0; | |
1522 | out: | |
1523 | btrfs_free_path(path); | |
ac8e9819 CM |
1524 | return ret; |
1525 | } | |
1526 | ||
1527 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, | |
1528 | void __user *argp) | |
1529 | { | |
1530 | struct btrfs_ioctl_ino_lookup_args *args; | |
1531 | struct inode *inode; | |
1532 | int ret; | |
1533 | ||
1534 | if (!capable(CAP_SYS_ADMIN)) | |
1535 | return -EPERM; | |
1536 | ||
2354d08f JL |
1537 | args = memdup_user(argp, sizeof(*args)); |
1538 | if (IS_ERR(args)) | |
1539 | return PTR_ERR(args); | |
c2b96929 | 1540 | |
ac8e9819 CM |
1541 | inode = fdentry(file)->d_inode; |
1542 | ||
1b53ac4d CM |
1543 | if (args->treeid == 0) |
1544 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
1545 | ||
ac8e9819 CM |
1546 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
1547 | args->treeid, args->objectid, | |
1548 | args->name); | |
1549 | ||
1550 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
1551 | ret = -EFAULT; | |
1552 | ||
1553 | kfree(args); | |
98d377a0 TH |
1554 | return ret; |
1555 | } | |
1556 | ||
76dda93c YZ |
1557 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
1558 | void __user *arg) | |
1559 | { | |
1560 | struct dentry *parent = fdentry(file); | |
1561 | struct dentry *dentry; | |
1562 | struct inode *dir = parent->d_inode; | |
1563 | struct inode *inode; | |
1564 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
1565 | struct btrfs_root *dest = NULL; | |
1566 | struct btrfs_ioctl_vol_args *vol_args; | |
1567 | struct btrfs_trans_handle *trans; | |
1568 | int namelen; | |
1569 | int ret; | |
1570 | int err = 0; | |
1571 | ||
76dda93c YZ |
1572 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1573 | if (IS_ERR(vol_args)) | |
1574 | return PTR_ERR(vol_args); | |
1575 | ||
1576 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
1577 | namelen = strlen(vol_args->name); | |
1578 | if (strchr(vol_args->name, '/') || | |
1579 | strncmp(vol_args->name, "..", namelen) == 0) { | |
1580 | err = -EINVAL; | |
1581 | goto out; | |
1582 | } | |
1583 | ||
1584 | err = mnt_want_write(file->f_path.mnt); | |
1585 | if (err) | |
1586 | goto out; | |
1587 | ||
1588 | mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT); | |
1589 | dentry = lookup_one_len(vol_args->name, parent, namelen); | |
1590 | if (IS_ERR(dentry)) { | |
1591 | err = PTR_ERR(dentry); | |
1592 | goto out_unlock_dir; | |
1593 | } | |
1594 | ||
1595 | if (!dentry->d_inode) { | |
1596 | err = -ENOENT; | |
1597 | goto out_dput; | |
1598 | } | |
1599 | ||
1600 | inode = dentry->d_inode; | |
4260f7c7 SW |
1601 | dest = BTRFS_I(inode)->root; |
1602 | if (!capable(CAP_SYS_ADMIN)){ | |
1603 | /* | |
1604 | * Regular user. Only allow this with a special mount | |
1605 | * option, when the user has write+exec access to the | |
1606 | * subvol root, and when rmdir(2) would have been | |
1607 | * allowed. | |
1608 | * | |
1609 | * Note that this is _not_ check that the subvol is | |
1610 | * empty or doesn't contain data that we wouldn't | |
1611 | * otherwise be able to delete. | |
1612 | * | |
1613 | * Users who want to delete empty subvols should try | |
1614 | * rmdir(2). | |
1615 | */ | |
1616 | err = -EPERM; | |
1617 | if (!btrfs_test_opt(root, USER_SUBVOL_RM_ALLOWED)) | |
1618 | goto out_dput; | |
1619 | ||
1620 | /* | |
1621 | * Do not allow deletion if the parent dir is the same | |
1622 | * as the dir to be deleted. That means the ioctl | |
1623 | * must be called on the dentry referencing the root | |
1624 | * of the subvol, not a random directory contained | |
1625 | * within it. | |
1626 | */ | |
1627 | err = -EINVAL; | |
1628 | if (root == dest) | |
1629 | goto out_dput; | |
1630 | ||
1631 | err = inode_permission(inode, MAY_WRITE | MAY_EXEC); | |
1632 | if (err) | |
1633 | goto out_dput; | |
1634 | ||
1635 | /* check if subvolume may be deleted by a non-root user */ | |
1636 | err = btrfs_may_delete(dir, dentry, 1); | |
1637 | if (err) | |
1638 | goto out_dput; | |
1639 | } | |
1640 | ||
33345d01 | 1641 | if (btrfs_ino(inode) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
1642 | err = -EINVAL; |
1643 | goto out_dput; | |
1644 | } | |
1645 | ||
76dda93c YZ |
1646 | mutex_lock(&inode->i_mutex); |
1647 | err = d_invalidate(dentry); | |
1648 | if (err) | |
1649 | goto out_unlock; | |
1650 | ||
1651 | down_write(&root->fs_info->subvol_sem); | |
1652 | ||
1653 | err = may_destroy_subvol(dest); | |
1654 | if (err) | |
1655 | goto out_up_write; | |
1656 | ||
a22285a6 YZ |
1657 | trans = btrfs_start_transaction(root, 0); |
1658 | if (IS_ERR(trans)) { | |
1659 | err = PTR_ERR(trans); | |
d327099a | 1660 | goto out_up_write; |
a22285a6 YZ |
1661 | } |
1662 | trans->block_rsv = &root->fs_info->global_block_rsv; | |
1663 | ||
76dda93c YZ |
1664 | ret = btrfs_unlink_subvol(trans, root, dir, |
1665 | dest->root_key.objectid, | |
1666 | dentry->d_name.name, | |
1667 | dentry->d_name.len); | |
1668 | BUG_ON(ret); | |
1669 | ||
1670 | btrfs_record_root_in_trans(trans, dest); | |
1671 | ||
1672 | memset(&dest->root_item.drop_progress, 0, | |
1673 | sizeof(dest->root_item.drop_progress)); | |
1674 | dest->root_item.drop_level = 0; | |
1675 | btrfs_set_root_refs(&dest->root_item, 0); | |
1676 | ||
d68fc57b YZ |
1677 | if (!xchg(&dest->orphan_item_inserted, 1)) { |
1678 | ret = btrfs_insert_orphan_item(trans, | |
1679 | root->fs_info->tree_root, | |
1680 | dest->root_key.objectid); | |
1681 | BUG_ON(ret); | |
1682 | } | |
76dda93c | 1683 | |
531cb13f | 1684 | ret = btrfs_end_transaction(trans, root); |
76dda93c YZ |
1685 | BUG_ON(ret); |
1686 | inode->i_flags |= S_DEAD; | |
1687 | out_up_write: | |
1688 | up_write(&root->fs_info->subvol_sem); | |
1689 | out_unlock: | |
1690 | mutex_unlock(&inode->i_mutex); | |
1691 | if (!err) { | |
efefb143 | 1692 | shrink_dcache_sb(root->fs_info->sb); |
76dda93c YZ |
1693 | btrfs_invalidate_inodes(dest); |
1694 | d_delete(dentry); | |
1695 | } | |
1696 | out_dput: | |
1697 | dput(dentry); | |
1698 | out_unlock_dir: | |
1699 | mutex_unlock(&dir->i_mutex); | |
1700 | mnt_drop_write(file->f_path.mnt); | |
1701 | out: | |
1702 | kfree(vol_args); | |
1703 | return err; | |
1704 | } | |
1705 | ||
1e701a32 | 1706 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 CH |
1707 | { |
1708 | struct inode *inode = fdentry(file)->d_inode; | |
1709 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
1e701a32 | 1710 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
1711 | int ret; |
1712 | ||
b83cc969 LZ |
1713 | if (btrfs_root_readonly(root)) |
1714 | return -EROFS; | |
1715 | ||
c146afad YZ |
1716 | ret = mnt_want_write(file->f_path.mnt); |
1717 | if (ret) | |
1718 | return ret; | |
f46b5a66 CH |
1719 | |
1720 | switch (inode->i_mode & S_IFMT) { | |
1721 | case S_IFDIR: | |
e441d54d CM |
1722 | if (!capable(CAP_SYS_ADMIN)) { |
1723 | ret = -EPERM; | |
1724 | goto out; | |
1725 | } | |
8929ecfa YZ |
1726 | ret = btrfs_defrag_root(root, 0); |
1727 | if (ret) | |
1728 | goto out; | |
1729 | ret = btrfs_defrag_root(root->fs_info->extent_root, 0); | |
f46b5a66 CH |
1730 | break; |
1731 | case S_IFREG: | |
e441d54d CM |
1732 | if (!(file->f_mode & FMODE_WRITE)) { |
1733 | ret = -EINVAL; | |
1734 | goto out; | |
1735 | } | |
1e701a32 CM |
1736 | |
1737 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
1738 | if (!range) { | |
1739 | ret = -ENOMEM; | |
1740 | goto out; | |
1741 | } | |
1742 | ||
1743 | if (argp) { | |
1744 | if (copy_from_user(range, argp, | |
1745 | sizeof(*range))) { | |
1746 | ret = -EFAULT; | |
1747 | kfree(range); | |
683be16e | 1748 | goto out; |
1e701a32 CM |
1749 | } |
1750 | /* compression requires us to start the IO */ | |
1751 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
1752 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
1753 | range->extent_thresh = (u32)-1; | |
1754 | } | |
1755 | } else { | |
1756 | /* the rest are all set to zero by kzalloc */ | |
1757 | range->len = (u64)-1; | |
1758 | } | |
8929ecfa | 1759 | ret = btrfs_defrag_file(file, range); |
1e701a32 | 1760 | kfree(range); |
f46b5a66 | 1761 | break; |
8929ecfa YZ |
1762 | default: |
1763 | ret = -EINVAL; | |
f46b5a66 | 1764 | } |
e441d54d | 1765 | out: |
ab67b7c1 | 1766 | mnt_drop_write(file->f_path.mnt); |
e441d54d | 1767 | return ret; |
f46b5a66 CH |
1768 | } |
1769 | ||
b2950863 | 1770 | static long btrfs_ioctl_add_dev(struct btrfs_root *root, void __user *arg) |
f46b5a66 CH |
1771 | { |
1772 | struct btrfs_ioctl_vol_args *vol_args; | |
1773 | int ret; | |
1774 | ||
e441d54d CM |
1775 | if (!capable(CAP_SYS_ADMIN)) |
1776 | return -EPERM; | |
1777 | ||
dae7b665 LZ |
1778 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1779 | if (IS_ERR(vol_args)) | |
1780 | return PTR_ERR(vol_args); | |
f46b5a66 | 1781 | |
5516e595 | 1782 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
f46b5a66 CH |
1783 | ret = btrfs_init_new_device(root, vol_args->name); |
1784 | ||
f46b5a66 CH |
1785 | kfree(vol_args); |
1786 | return ret; | |
1787 | } | |
1788 | ||
b2950863 | 1789 | static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg) |
f46b5a66 CH |
1790 | { |
1791 | struct btrfs_ioctl_vol_args *vol_args; | |
1792 | int ret; | |
1793 | ||
e441d54d CM |
1794 | if (!capable(CAP_SYS_ADMIN)) |
1795 | return -EPERM; | |
1796 | ||
c146afad YZ |
1797 | if (root->fs_info->sb->s_flags & MS_RDONLY) |
1798 | return -EROFS; | |
1799 | ||
dae7b665 LZ |
1800 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1801 | if (IS_ERR(vol_args)) | |
1802 | return PTR_ERR(vol_args); | |
f46b5a66 | 1803 | |
5516e595 | 1804 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
f46b5a66 CH |
1805 | ret = btrfs_rm_device(root, vol_args->name); |
1806 | ||
f46b5a66 CH |
1807 | kfree(vol_args); |
1808 | return ret; | |
1809 | } | |
1810 | ||
475f6387 JS |
1811 | static long btrfs_ioctl_fs_info(struct btrfs_root *root, void __user *arg) |
1812 | { | |
1813 | struct btrfs_ioctl_fs_info_args fi_args; | |
1814 | struct btrfs_device *device; | |
1815 | struct btrfs_device *next; | |
1816 | struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; | |
1817 | ||
1818 | if (!capable(CAP_SYS_ADMIN)) | |
1819 | return -EPERM; | |
1820 | ||
1821 | fi_args.num_devices = fs_devices->num_devices; | |
1822 | fi_args.max_id = 0; | |
1823 | memcpy(&fi_args.fsid, root->fs_info->fsid, sizeof(fi_args.fsid)); | |
1824 | ||
1825 | mutex_lock(&fs_devices->device_list_mutex); | |
1826 | list_for_each_entry_safe(device, next, &fs_devices->devices, dev_list) { | |
1827 | if (device->devid > fi_args.max_id) | |
1828 | fi_args.max_id = device->devid; | |
1829 | } | |
1830 | mutex_unlock(&fs_devices->device_list_mutex); | |
1831 | ||
1832 | if (copy_to_user(arg, &fi_args, sizeof(fi_args))) | |
1833 | return -EFAULT; | |
1834 | ||
1835 | return 0; | |
1836 | } | |
1837 | ||
1838 | static long btrfs_ioctl_dev_info(struct btrfs_root *root, void __user *arg) | |
1839 | { | |
1840 | struct btrfs_ioctl_dev_info_args *di_args; | |
1841 | struct btrfs_device *dev; | |
1842 | struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices; | |
1843 | int ret = 0; | |
1844 | char *s_uuid = NULL; | |
1845 | char empty_uuid[BTRFS_UUID_SIZE] = {0}; | |
1846 | ||
1847 | if (!capable(CAP_SYS_ADMIN)) | |
1848 | return -EPERM; | |
1849 | ||
1850 | di_args = memdup_user(arg, sizeof(*di_args)); | |
1851 | if (IS_ERR(di_args)) | |
1852 | return PTR_ERR(di_args); | |
1853 | ||
1854 | if (memcmp(empty_uuid, di_args->uuid, BTRFS_UUID_SIZE) != 0) | |
1855 | s_uuid = di_args->uuid; | |
1856 | ||
1857 | mutex_lock(&fs_devices->device_list_mutex); | |
1858 | dev = btrfs_find_device(root, di_args->devid, s_uuid, NULL); | |
1859 | mutex_unlock(&fs_devices->device_list_mutex); | |
1860 | ||
1861 | if (!dev) { | |
1862 | ret = -ENODEV; | |
1863 | goto out; | |
1864 | } | |
1865 | ||
1866 | di_args->devid = dev->devid; | |
1867 | di_args->bytes_used = dev->bytes_used; | |
1868 | di_args->total_bytes = dev->total_bytes; | |
1869 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); | |
1870 | strncpy(di_args->path, dev->name, sizeof(di_args->path)); | |
1871 | ||
1872 | out: | |
1873 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) | |
1874 | ret = -EFAULT; | |
1875 | ||
1876 | kfree(di_args); | |
1877 | return ret; | |
1878 | } | |
1879 | ||
76dda93c YZ |
1880 | static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd, |
1881 | u64 off, u64 olen, u64 destoff) | |
f46b5a66 CH |
1882 | { |
1883 | struct inode *inode = fdentry(file)->d_inode; | |
1884 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
1885 | struct file *src_file; | |
1886 | struct inode *src; | |
1887 | struct btrfs_trans_handle *trans; | |
f46b5a66 | 1888 | struct btrfs_path *path; |
f46b5a66 | 1889 | struct extent_buffer *leaf; |
ae01a0ab YZ |
1890 | char *buf; |
1891 | struct btrfs_key key; | |
f46b5a66 CH |
1892 | u32 nritems; |
1893 | int slot; | |
ae01a0ab | 1894 | int ret; |
c5c9cd4d SW |
1895 | u64 len = olen; |
1896 | u64 bs = root->fs_info->sb->s_blocksize; | |
1897 | u64 hint_byte; | |
d20f7043 | 1898 | |
c5c9cd4d SW |
1899 | /* |
1900 | * TODO: | |
1901 | * - split compressed inline extents. annoying: we need to | |
1902 | * decompress into destination's address_space (the file offset | |
1903 | * may change, so source mapping won't do), then recompress (or | |
1904 | * otherwise reinsert) a subrange. | |
1905 | * - allow ranges within the same file to be cloned (provided | |
1906 | * they don't overlap)? | |
1907 | */ | |
1908 | ||
e441d54d | 1909 | /* the destination must be opened for writing */ |
2ebc3464 | 1910 | if (!(file->f_mode & FMODE_WRITE) || (file->f_flags & O_APPEND)) |
e441d54d CM |
1911 | return -EINVAL; |
1912 | ||
b83cc969 LZ |
1913 | if (btrfs_root_readonly(root)) |
1914 | return -EROFS; | |
1915 | ||
c146afad YZ |
1916 | ret = mnt_want_write(file->f_path.mnt); |
1917 | if (ret) | |
1918 | return ret; | |
1919 | ||
c5c9cd4d | 1920 | src_file = fget(srcfd); |
ab67b7c1 YZ |
1921 | if (!src_file) { |
1922 | ret = -EBADF; | |
1923 | goto out_drop_write; | |
1924 | } | |
5dc64164 | 1925 | |
f46b5a66 CH |
1926 | src = src_file->f_dentry->d_inode; |
1927 | ||
c5c9cd4d SW |
1928 | ret = -EINVAL; |
1929 | if (src == inode) | |
1930 | goto out_fput; | |
1931 | ||
5dc64164 DR |
1932 | /* the src must be open for reading */ |
1933 | if (!(src_file->f_mode & FMODE_READ)) | |
1934 | goto out_fput; | |
1935 | ||
ae01a0ab YZ |
1936 | ret = -EISDIR; |
1937 | if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode)) | |
1938 | goto out_fput; | |
1939 | ||
f46b5a66 | 1940 | ret = -EXDEV; |
ae01a0ab YZ |
1941 | if (src->i_sb != inode->i_sb || BTRFS_I(src)->root != root) |
1942 | goto out_fput; | |
1943 | ||
1944 | ret = -ENOMEM; | |
1945 | buf = vmalloc(btrfs_level_size(root, 0)); | |
1946 | if (!buf) | |
1947 | goto out_fput; | |
1948 | ||
1949 | path = btrfs_alloc_path(); | |
1950 | if (!path) { | |
1951 | vfree(buf); | |
f46b5a66 | 1952 | goto out_fput; |
ae01a0ab YZ |
1953 | } |
1954 | path->reada = 2; | |
f46b5a66 CH |
1955 | |
1956 | if (inode < src) { | |
fccdae43 SW |
1957 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT); |
1958 | mutex_lock_nested(&src->i_mutex, I_MUTEX_CHILD); | |
f46b5a66 | 1959 | } else { |
fccdae43 SW |
1960 | mutex_lock_nested(&src->i_mutex, I_MUTEX_PARENT); |
1961 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_CHILD); | |
f46b5a66 CH |
1962 | } |
1963 | ||
c5c9cd4d SW |
1964 | /* determine range to clone */ |
1965 | ret = -EINVAL; | |
2ebc3464 | 1966 | if (off + len > src->i_size || off + len < off) |
f46b5a66 | 1967 | goto out_unlock; |
c5c9cd4d SW |
1968 | if (len == 0) |
1969 | olen = len = src->i_size - off; | |
1970 | /* if we extend to eof, continue to block boundary */ | |
1971 | if (off + len == src->i_size) | |
2a6b8dae | 1972 | len = ALIGN(src->i_size, bs) - off; |
c5c9cd4d SW |
1973 | |
1974 | /* verify the end result is block aligned */ | |
2a6b8dae LZ |
1975 | if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) || |
1976 | !IS_ALIGNED(destoff, bs)) | |
c5c9cd4d SW |
1977 | goto out_unlock; |
1978 | ||
f46b5a66 CH |
1979 | /* do any pending delalloc/csum calc on src, one way or |
1980 | another, and lock file content */ | |
1981 | while (1) { | |
31840ae1 | 1982 | struct btrfs_ordered_extent *ordered; |
c5c9cd4d | 1983 | lock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS); |
9a019196 SW |
1984 | ordered = btrfs_lookup_first_ordered_extent(src, off+len); |
1985 | if (!ordered && | |
1986 | !test_range_bit(&BTRFS_I(src)->io_tree, off, off+len, | |
1987 | EXTENT_DELALLOC, 0, NULL)) | |
f46b5a66 | 1988 | break; |
c5c9cd4d | 1989 | unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS); |
ae01a0ab YZ |
1990 | if (ordered) |
1991 | btrfs_put_ordered_extent(ordered); | |
9a019196 | 1992 | btrfs_wait_ordered_range(src, off, len); |
f46b5a66 CH |
1993 | } |
1994 | ||
c5c9cd4d | 1995 | /* clone data */ |
33345d01 | 1996 | key.objectid = btrfs_ino(src); |
ae01a0ab YZ |
1997 | key.type = BTRFS_EXTENT_DATA_KEY; |
1998 | key.offset = 0; | |
f46b5a66 CH |
1999 | |
2000 | while (1) { | |
2001 | /* | |
2002 | * note the key will change type as we walk through the | |
2003 | * tree. | |
2004 | */ | |
a22285a6 | 2005 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
f46b5a66 CH |
2006 | if (ret < 0) |
2007 | goto out; | |
2008 | ||
ae01a0ab YZ |
2009 | nritems = btrfs_header_nritems(path->nodes[0]); |
2010 | if (path->slots[0] >= nritems) { | |
f46b5a66 CH |
2011 | ret = btrfs_next_leaf(root, path); |
2012 | if (ret < 0) | |
2013 | goto out; | |
2014 | if (ret > 0) | |
2015 | break; | |
ae01a0ab | 2016 | nritems = btrfs_header_nritems(path->nodes[0]); |
f46b5a66 CH |
2017 | } |
2018 | leaf = path->nodes[0]; | |
2019 | slot = path->slots[0]; | |
f46b5a66 | 2020 | |
ae01a0ab | 2021 | btrfs_item_key_to_cpu(leaf, &key, slot); |
d20f7043 | 2022 | if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY || |
33345d01 | 2023 | key.objectid != btrfs_ino(src)) |
f46b5a66 CH |
2024 | break; |
2025 | ||
c5c9cd4d SW |
2026 | if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) { |
2027 | struct btrfs_file_extent_item *extent; | |
2028 | int type; | |
31840ae1 ZY |
2029 | u32 size; |
2030 | struct btrfs_key new_key; | |
c5c9cd4d SW |
2031 | u64 disko = 0, diskl = 0; |
2032 | u64 datao = 0, datal = 0; | |
2033 | u8 comp; | |
b5384d48 | 2034 | u64 endoff; |
31840ae1 ZY |
2035 | |
2036 | size = btrfs_item_size_nr(leaf, slot); | |
2037 | read_extent_buffer(leaf, buf, | |
2038 | btrfs_item_ptr_offset(leaf, slot), | |
2039 | size); | |
c5c9cd4d SW |
2040 | |
2041 | extent = btrfs_item_ptr(leaf, slot, | |
2042 | struct btrfs_file_extent_item); | |
2043 | comp = btrfs_file_extent_compression(leaf, extent); | |
2044 | type = btrfs_file_extent_type(leaf, extent); | |
c8a894d7 CM |
2045 | if (type == BTRFS_FILE_EXTENT_REG || |
2046 | type == BTRFS_FILE_EXTENT_PREALLOC) { | |
d397712b CM |
2047 | disko = btrfs_file_extent_disk_bytenr(leaf, |
2048 | extent); | |
2049 | diskl = btrfs_file_extent_disk_num_bytes(leaf, | |
2050 | extent); | |
c5c9cd4d | 2051 | datao = btrfs_file_extent_offset(leaf, extent); |
d397712b CM |
2052 | datal = btrfs_file_extent_num_bytes(leaf, |
2053 | extent); | |
c5c9cd4d SW |
2054 | } else if (type == BTRFS_FILE_EXTENT_INLINE) { |
2055 | /* take upper bound, may be compressed */ | |
2056 | datal = btrfs_file_extent_ram_bytes(leaf, | |
2057 | extent); | |
2058 | } | |
b3b4aa74 | 2059 | btrfs_release_path(path); |
31840ae1 | 2060 | |
050006a7 | 2061 | if (key.offset + datal <= off || |
c5c9cd4d SW |
2062 | key.offset >= off+len) |
2063 | goto next; | |
2064 | ||
31840ae1 | 2065 | memcpy(&new_key, &key, sizeof(new_key)); |
33345d01 | 2066 | new_key.objectid = btrfs_ino(inode); |
4d728ec7 LZ |
2067 | if (off <= key.offset) |
2068 | new_key.offset = key.offset + destoff - off; | |
2069 | else | |
2070 | new_key.offset = destoff; | |
31840ae1 | 2071 | |
a22285a6 YZ |
2072 | trans = btrfs_start_transaction(root, 1); |
2073 | if (IS_ERR(trans)) { | |
2074 | ret = PTR_ERR(trans); | |
2075 | goto out; | |
2076 | } | |
2077 | ||
c8a894d7 CM |
2078 | if (type == BTRFS_FILE_EXTENT_REG || |
2079 | type == BTRFS_FILE_EXTENT_PREALLOC) { | |
a22285a6 YZ |
2080 | if (off > key.offset) { |
2081 | datao += off - key.offset; | |
2082 | datal -= off - key.offset; | |
2083 | } | |
2084 | ||
2085 | if (key.offset + datal > off + len) | |
2086 | datal = off + len - key.offset; | |
2087 | ||
2088 | ret = btrfs_drop_extents(trans, inode, | |
2089 | new_key.offset, | |
2090 | new_key.offset + datal, | |
2091 | &hint_byte, 1); | |
2092 | BUG_ON(ret); | |
2093 | ||
c5c9cd4d SW |
2094 | ret = btrfs_insert_empty_item(trans, root, path, |
2095 | &new_key, size); | |
a22285a6 | 2096 | BUG_ON(ret); |
c5c9cd4d SW |
2097 | |
2098 | leaf = path->nodes[0]; | |
2099 | slot = path->slots[0]; | |
2100 | write_extent_buffer(leaf, buf, | |
31840ae1 ZY |
2101 | btrfs_item_ptr_offset(leaf, slot), |
2102 | size); | |
ae01a0ab | 2103 | |
c5c9cd4d | 2104 | extent = btrfs_item_ptr(leaf, slot, |
f46b5a66 | 2105 | struct btrfs_file_extent_item); |
c5c9cd4d | 2106 | |
c5c9cd4d SW |
2107 | /* disko == 0 means it's a hole */ |
2108 | if (!disko) | |
2109 | datao = 0; | |
c5c9cd4d SW |
2110 | |
2111 | btrfs_set_file_extent_offset(leaf, extent, | |
2112 | datao); | |
2113 | btrfs_set_file_extent_num_bytes(leaf, extent, | |
2114 | datal); | |
2115 | if (disko) { | |
2116 | inode_add_bytes(inode, datal); | |
ae01a0ab | 2117 | ret = btrfs_inc_extent_ref(trans, root, |
5d4f98a2 YZ |
2118 | disko, diskl, 0, |
2119 | root->root_key.objectid, | |
33345d01 | 2120 | btrfs_ino(inode), |
5d4f98a2 | 2121 | new_key.offset - datao); |
31840ae1 | 2122 | BUG_ON(ret); |
f46b5a66 | 2123 | } |
c5c9cd4d SW |
2124 | } else if (type == BTRFS_FILE_EXTENT_INLINE) { |
2125 | u64 skip = 0; | |
2126 | u64 trim = 0; | |
2127 | if (off > key.offset) { | |
2128 | skip = off - key.offset; | |
2129 | new_key.offset += skip; | |
2130 | } | |
d397712b | 2131 | |
c5c9cd4d SW |
2132 | if (key.offset + datal > off+len) |
2133 | trim = key.offset + datal - (off+len); | |
d397712b | 2134 | |
c5c9cd4d | 2135 | if (comp && (skip || trim)) { |
c5c9cd4d | 2136 | ret = -EINVAL; |
a22285a6 | 2137 | btrfs_end_transaction(trans, root); |
c5c9cd4d SW |
2138 | goto out; |
2139 | } | |
2140 | size -= skip + trim; | |
2141 | datal -= skip + trim; | |
a22285a6 YZ |
2142 | |
2143 | ret = btrfs_drop_extents(trans, inode, | |
2144 | new_key.offset, | |
2145 | new_key.offset + datal, | |
2146 | &hint_byte, 1); | |
2147 | BUG_ON(ret); | |
2148 | ||
c5c9cd4d SW |
2149 | ret = btrfs_insert_empty_item(trans, root, path, |
2150 | &new_key, size); | |
a22285a6 | 2151 | BUG_ON(ret); |
c5c9cd4d SW |
2152 | |
2153 | if (skip) { | |
d397712b CM |
2154 | u32 start = |
2155 | btrfs_file_extent_calc_inline_size(0); | |
c5c9cd4d SW |
2156 | memmove(buf+start, buf+start+skip, |
2157 | datal); | |
2158 | } | |
2159 | ||
2160 | leaf = path->nodes[0]; | |
2161 | slot = path->slots[0]; | |
2162 | write_extent_buffer(leaf, buf, | |
2163 | btrfs_item_ptr_offset(leaf, slot), | |
2164 | size); | |
2165 | inode_add_bytes(inode, datal); | |
f46b5a66 | 2166 | } |
c5c9cd4d SW |
2167 | |
2168 | btrfs_mark_buffer_dirty(leaf); | |
b3b4aa74 | 2169 | btrfs_release_path(path); |
c5c9cd4d | 2170 | |
a22285a6 | 2171 | inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
b5384d48 SW |
2172 | |
2173 | /* | |
2174 | * we round up to the block size at eof when | |
2175 | * determining which extents to clone above, | |
2176 | * but shouldn't round up the file size | |
2177 | */ | |
2178 | endoff = new_key.offset + datal; | |
5f3888ff LZ |
2179 | if (endoff > destoff+olen) |
2180 | endoff = destoff+olen; | |
b5384d48 SW |
2181 | if (endoff > inode->i_size) |
2182 | btrfs_i_size_write(inode, endoff); | |
2183 | ||
a22285a6 YZ |
2184 | BTRFS_I(inode)->flags = BTRFS_I(src)->flags; |
2185 | ret = btrfs_update_inode(trans, root, inode); | |
2186 | BUG_ON(ret); | |
2187 | btrfs_end_transaction(trans, root); | |
2188 | } | |
d397712b | 2189 | next: |
b3b4aa74 | 2190 | btrfs_release_path(path); |
f46b5a66 | 2191 | key.offset++; |
f46b5a66 | 2192 | } |
f46b5a66 CH |
2193 | ret = 0; |
2194 | out: | |
b3b4aa74 | 2195 | btrfs_release_path(path); |
c5c9cd4d | 2196 | unlock_extent(&BTRFS_I(src)->io_tree, off, off+len, GFP_NOFS); |
f46b5a66 CH |
2197 | out_unlock: |
2198 | mutex_unlock(&src->i_mutex); | |
2199 | mutex_unlock(&inode->i_mutex); | |
ae01a0ab YZ |
2200 | vfree(buf); |
2201 | btrfs_free_path(path); | |
f46b5a66 CH |
2202 | out_fput: |
2203 | fput(src_file); | |
ab67b7c1 YZ |
2204 | out_drop_write: |
2205 | mnt_drop_write(file->f_path.mnt); | |
f46b5a66 CH |
2206 | return ret; |
2207 | } | |
2208 | ||
7a865e8a | 2209 | static long btrfs_ioctl_clone_range(struct file *file, void __user *argp) |
c5c9cd4d SW |
2210 | { |
2211 | struct btrfs_ioctl_clone_range_args args; | |
2212 | ||
7a865e8a | 2213 | if (copy_from_user(&args, argp, sizeof(args))) |
c5c9cd4d SW |
2214 | return -EFAULT; |
2215 | return btrfs_ioctl_clone(file, args.src_fd, args.src_offset, | |
2216 | args.src_length, args.dest_offset); | |
2217 | } | |
2218 | ||
f46b5a66 CH |
2219 | /* |
2220 | * there are many ways the trans_start and trans_end ioctls can lead | |
2221 | * to deadlocks. They should only be used by applications that | |
2222 | * basically own the machine, and have a very in depth understanding | |
2223 | * of all the possible deadlocks and enospc problems. | |
2224 | */ | |
b2950863 | 2225 | static long btrfs_ioctl_trans_start(struct file *file) |
f46b5a66 CH |
2226 | { |
2227 | struct inode *inode = fdentry(file)->d_inode; | |
2228 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
2229 | struct btrfs_trans_handle *trans; | |
1ab86aed | 2230 | int ret; |
f46b5a66 | 2231 | |
1ab86aed | 2232 | ret = -EPERM; |
df5b5520 | 2233 | if (!capable(CAP_SYS_ADMIN)) |
1ab86aed | 2234 | goto out; |
df5b5520 | 2235 | |
1ab86aed SW |
2236 | ret = -EINPROGRESS; |
2237 | if (file->private_data) | |
f46b5a66 | 2238 | goto out; |
9ca9ee09 | 2239 | |
b83cc969 LZ |
2240 | ret = -EROFS; |
2241 | if (btrfs_root_readonly(root)) | |
2242 | goto out; | |
2243 | ||
c146afad YZ |
2244 | ret = mnt_want_write(file->f_path.mnt); |
2245 | if (ret) | |
2246 | goto out; | |
2247 | ||
9ca9ee09 SW |
2248 | mutex_lock(&root->fs_info->trans_mutex); |
2249 | root->fs_info->open_ioctl_trans++; | |
2250 | mutex_unlock(&root->fs_info->trans_mutex); | |
2251 | ||
1ab86aed | 2252 | ret = -ENOMEM; |
9ca9ee09 | 2253 | trans = btrfs_start_ioctl_transaction(root, 0); |
abd30bb0 | 2254 | if (IS_ERR(trans)) |
1ab86aed SW |
2255 | goto out_drop; |
2256 | ||
2257 | file->private_data = trans; | |
2258 | return 0; | |
2259 | ||
2260 | out_drop: | |
2261 | mutex_lock(&root->fs_info->trans_mutex); | |
2262 | root->fs_info->open_ioctl_trans--; | |
2263 | mutex_unlock(&root->fs_info->trans_mutex); | |
2264 | mnt_drop_write(file->f_path.mnt); | |
f46b5a66 | 2265 | out: |
f46b5a66 CH |
2266 | return ret; |
2267 | } | |
2268 | ||
6ef5ed0d JB |
2269 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
2270 | { | |
2271 | struct inode *inode = fdentry(file)->d_inode; | |
2272 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
2273 | struct btrfs_root *new_root; | |
2274 | struct btrfs_dir_item *di; | |
2275 | struct btrfs_trans_handle *trans; | |
2276 | struct btrfs_path *path; | |
2277 | struct btrfs_key location; | |
2278 | struct btrfs_disk_key disk_key; | |
2279 | struct btrfs_super_block *disk_super; | |
2280 | u64 features; | |
2281 | u64 objectid = 0; | |
2282 | u64 dir_id; | |
2283 | ||
2284 | if (!capable(CAP_SYS_ADMIN)) | |
2285 | return -EPERM; | |
2286 | ||
2287 | if (copy_from_user(&objectid, argp, sizeof(objectid))) | |
2288 | return -EFAULT; | |
2289 | ||
2290 | if (!objectid) | |
2291 | objectid = root->root_key.objectid; | |
2292 | ||
2293 | location.objectid = objectid; | |
2294 | location.type = BTRFS_ROOT_ITEM_KEY; | |
2295 | location.offset = (u64)-1; | |
2296 | ||
2297 | new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); | |
2298 | if (IS_ERR(new_root)) | |
2299 | return PTR_ERR(new_root); | |
2300 | ||
2301 | if (btrfs_root_refs(&new_root->root_item) == 0) | |
2302 | return -ENOENT; | |
2303 | ||
2304 | path = btrfs_alloc_path(); | |
2305 | if (!path) | |
2306 | return -ENOMEM; | |
2307 | path->leave_spinning = 1; | |
2308 | ||
2309 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 2310 | if (IS_ERR(trans)) { |
6ef5ed0d | 2311 | btrfs_free_path(path); |
98d5dc13 | 2312 | return PTR_ERR(trans); |
6ef5ed0d JB |
2313 | } |
2314 | ||
2315 | dir_id = btrfs_super_root_dir(&root->fs_info->super_copy); | |
2316 | di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path, | |
2317 | dir_id, "default", 7, 1); | |
cf1e99a4 | 2318 | if (IS_ERR_OR_NULL(di)) { |
6ef5ed0d JB |
2319 | btrfs_free_path(path); |
2320 | btrfs_end_transaction(trans, root); | |
2321 | printk(KERN_ERR "Umm, you don't have the default dir item, " | |
2322 | "this isn't going to work\n"); | |
2323 | return -ENOENT; | |
2324 | } | |
2325 | ||
2326 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
2327 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
2328 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2329 | btrfs_free_path(path); | |
2330 | ||
2331 | disk_super = &root->fs_info->super_copy; | |
2332 | features = btrfs_super_incompat_flags(disk_super); | |
2333 | if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) { | |
2334 | features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL; | |
2335 | btrfs_set_super_incompat_flags(disk_super, features); | |
2336 | } | |
2337 | btrfs_end_transaction(trans, root); | |
2338 | ||
2339 | return 0; | |
2340 | } | |
2341 | ||
bf5fc093 JB |
2342 | static void get_block_group_info(struct list_head *groups_list, |
2343 | struct btrfs_ioctl_space_info *space) | |
2344 | { | |
2345 | struct btrfs_block_group_cache *block_group; | |
2346 | ||
2347 | space->total_bytes = 0; | |
2348 | space->used_bytes = 0; | |
2349 | space->flags = 0; | |
2350 | list_for_each_entry(block_group, groups_list, list) { | |
2351 | space->flags = block_group->flags; | |
2352 | space->total_bytes += block_group->key.offset; | |
2353 | space->used_bytes += | |
2354 | btrfs_block_group_used(&block_group->item); | |
2355 | } | |
2356 | } | |
2357 | ||
1406e432 JB |
2358 | long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg) |
2359 | { | |
2360 | struct btrfs_ioctl_space_args space_args; | |
2361 | struct btrfs_ioctl_space_info space; | |
2362 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 2363 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 2364 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 2365 | struct btrfs_space_info *info; |
bf5fc093 JB |
2366 | u64 types[] = {BTRFS_BLOCK_GROUP_DATA, |
2367 | BTRFS_BLOCK_GROUP_SYSTEM, | |
2368 | BTRFS_BLOCK_GROUP_METADATA, | |
2369 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA}; | |
2370 | int num_types = 4; | |
7fde62bf | 2371 | int alloc_size; |
1406e432 | 2372 | int ret = 0; |
51788b1b | 2373 | u64 slot_count = 0; |
bf5fc093 | 2374 | int i, c; |
1406e432 JB |
2375 | |
2376 | if (copy_from_user(&space_args, | |
2377 | (struct btrfs_ioctl_space_args __user *)arg, | |
2378 | sizeof(space_args))) | |
2379 | return -EFAULT; | |
2380 | ||
bf5fc093 JB |
2381 | for (i = 0; i < num_types; i++) { |
2382 | struct btrfs_space_info *tmp; | |
2383 | ||
2384 | info = NULL; | |
2385 | rcu_read_lock(); | |
2386 | list_for_each_entry_rcu(tmp, &root->fs_info->space_info, | |
2387 | list) { | |
2388 | if (tmp->flags == types[i]) { | |
2389 | info = tmp; | |
2390 | break; | |
2391 | } | |
2392 | } | |
2393 | rcu_read_unlock(); | |
2394 | ||
2395 | if (!info) | |
2396 | continue; | |
2397 | ||
2398 | down_read(&info->groups_sem); | |
2399 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
2400 | if (!list_empty(&info->block_groups[c])) | |
2401 | slot_count++; | |
2402 | } | |
2403 | up_read(&info->groups_sem); | |
2404 | } | |
7fde62bf CM |
2405 | |
2406 | /* space_slots == 0 means they are asking for a count */ | |
2407 | if (space_args.space_slots == 0) { | |
2408 | space_args.total_spaces = slot_count; | |
2409 | goto out; | |
2410 | } | |
bf5fc093 | 2411 | |
51788b1b | 2412 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 2413 | |
7fde62bf | 2414 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 2415 | |
7fde62bf CM |
2416 | /* we generally have at most 6 or so space infos, one for each raid |
2417 | * level. So, a whole page should be more than enough for everyone | |
2418 | */ | |
2419 | if (alloc_size > PAGE_CACHE_SIZE) | |
2420 | return -ENOMEM; | |
2421 | ||
1406e432 | 2422 | space_args.total_spaces = 0; |
7fde62bf CM |
2423 | dest = kmalloc(alloc_size, GFP_NOFS); |
2424 | if (!dest) | |
2425 | return -ENOMEM; | |
2426 | dest_orig = dest; | |
1406e432 | 2427 | |
7fde62bf | 2428 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
2429 | for (i = 0; i < num_types; i++) { |
2430 | struct btrfs_space_info *tmp; | |
2431 | ||
51788b1b DR |
2432 | if (!slot_count) |
2433 | break; | |
2434 | ||
bf5fc093 JB |
2435 | info = NULL; |
2436 | rcu_read_lock(); | |
2437 | list_for_each_entry_rcu(tmp, &root->fs_info->space_info, | |
2438 | list) { | |
2439 | if (tmp->flags == types[i]) { | |
2440 | info = tmp; | |
2441 | break; | |
2442 | } | |
2443 | } | |
2444 | rcu_read_unlock(); | |
7fde62bf | 2445 | |
bf5fc093 JB |
2446 | if (!info) |
2447 | continue; | |
2448 | down_read(&info->groups_sem); | |
2449 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
2450 | if (!list_empty(&info->block_groups[c])) { | |
2451 | get_block_group_info(&info->block_groups[c], | |
2452 | &space); | |
2453 | memcpy(dest, &space, sizeof(space)); | |
2454 | dest++; | |
2455 | space_args.total_spaces++; | |
51788b1b | 2456 | slot_count--; |
bf5fc093 | 2457 | } |
51788b1b DR |
2458 | if (!slot_count) |
2459 | break; | |
bf5fc093 JB |
2460 | } |
2461 | up_read(&info->groups_sem); | |
1406e432 | 2462 | } |
1406e432 | 2463 | |
7fde62bf CM |
2464 | user_dest = (struct btrfs_ioctl_space_info *) |
2465 | (arg + sizeof(struct btrfs_ioctl_space_args)); | |
2466 | ||
2467 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
2468 | ret = -EFAULT; | |
2469 | ||
2470 | kfree(dest_orig); | |
2471 | out: | |
2472 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
2473 | ret = -EFAULT; |
2474 | ||
2475 | return ret; | |
2476 | } | |
2477 | ||
f46b5a66 CH |
2478 | /* |
2479 | * there are many ways the trans_start and trans_end ioctls can lead | |
2480 | * to deadlocks. They should only be used by applications that | |
2481 | * basically own the machine, and have a very in depth understanding | |
2482 | * of all the possible deadlocks and enospc problems. | |
2483 | */ | |
2484 | long btrfs_ioctl_trans_end(struct file *file) | |
2485 | { | |
2486 | struct inode *inode = fdentry(file)->d_inode; | |
2487 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
2488 | struct btrfs_trans_handle *trans; | |
f46b5a66 | 2489 | |
f46b5a66 | 2490 | trans = file->private_data; |
1ab86aed SW |
2491 | if (!trans) |
2492 | return -EINVAL; | |
b214107e | 2493 | file->private_data = NULL; |
9ca9ee09 | 2494 | |
1ab86aed SW |
2495 | btrfs_end_transaction(trans, root); |
2496 | ||
9ca9ee09 SW |
2497 | mutex_lock(&root->fs_info->trans_mutex); |
2498 | root->fs_info->open_ioctl_trans--; | |
2499 | mutex_unlock(&root->fs_info->trans_mutex); | |
2500 | ||
cfc8ea87 | 2501 | mnt_drop_write(file->f_path.mnt); |
1ab86aed | 2502 | return 0; |
f46b5a66 CH |
2503 | } |
2504 | ||
46204592 SW |
2505 | static noinline long btrfs_ioctl_start_sync(struct file *file, void __user *argp) |
2506 | { | |
2507 | struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root; | |
2508 | struct btrfs_trans_handle *trans; | |
2509 | u64 transid; | |
db5b493a | 2510 | int ret; |
46204592 SW |
2511 | |
2512 | trans = btrfs_start_transaction(root, 0); | |
98d5dc13 TI |
2513 | if (IS_ERR(trans)) |
2514 | return PTR_ERR(trans); | |
46204592 | 2515 | transid = trans->transid; |
db5b493a | 2516 | ret = btrfs_commit_transaction_async(trans, root, 0); |
8b2b2d3c TI |
2517 | if (ret) { |
2518 | btrfs_end_transaction(trans, root); | |
db5b493a | 2519 | return ret; |
8b2b2d3c | 2520 | } |
46204592 SW |
2521 | |
2522 | if (argp) | |
2523 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
2524 | return -EFAULT; | |
2525 | return 0; | |
2526 | } | |
2527 | ||
2528 | static noinline long btrfs_ioctl_wait_sync(struct file *file, void __user *argp) | |
2529 | { | |
2530 | struct btrfs_root *root = BTRFS_I(file->f_dentry->d_inode)->root; | |
2531 | u64 transid; | |
2532 | ||
2533 | if (argp) { | |
2534 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
2535 | return -EFAULT; | |
2536 | } else { | |
2537 | transid = 0; /* current trans */ | |
2538 | } | |
2539 | return btrfs_wait_for_commit(root, transid); | |
2540 | } | |
2541 | ||
475f6387 JS |
2542 | static long btrfs_ioctl_scrub(struct btrfs_root *root, void __user *arg) |
2543 | { | |
2544 | int ret; | |
2545 | struct btrfs_ioctl_scrub_args *sa; | |
2546 | ||
2547 | if (!capable(CAP_SYS_ADMIN)) | |
2548 | return -EPERM; | |
2549 | ||
2550 | sa = memdup_user(arg, sizeof(*sa)); | |
2551 | if (IS_ERR(sa)) | |
2552 | return PTR_ERR(sa); | |
2553 | ||
2554 | ret = btrfs_scrub_dev(root, sa->devid, sa->start, sa->end, | |
8628764e | 2555 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY); |
475f6387 JS |
2556 | |
2557 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
2558 | ret = -EFAULT; | |
2559 | ||
2560 | kfree(sa); | |
2561 | return ret; | |
2562 | } | |
2563 | ||
2564 | static long btrfs_ioctl_scrub_cancel(struct btrfs_root *root, void __user *arg) | |
2565 | { | |
2566 | if (!capable(CAP_SYS_ADMIN)) | |
2567 | return -EPERM; | |
2568 | ||
2569 | return btrfs_scrub_cancel(root); | |
2570 | } | |
2571 | ||
2572 | static long btrfs_ioctl_scrub_progress(struct btrfs_root *root, | |
2573 | void __user *arg) | |
2574 | { | |
2575 | struct btrfs_ioctl_scrub_args *sa; | |
2576 | int ret; | |
2577 | ||
2578 | if (!capable(CAP_SYS_ADMIN)) | |
2579 | return -EPERM; | |
2580 | ||
2581 | sa = memdup_user(arg, sizeof(*sa)); | |
2582 | if (IS_ERR(sa)) | |
2583 | return PTR_ERR(sa); | |
2584 | ||
2585 | ret = btrfs_scrub_progress(root, sa->devid, &sa->progress); | |
2586 | ||
2587 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
2588 | ret = -EFAULT; | |
2589 | ||
2590 | kfree(sa); | |
2591 | return ret; | |
2592 | } | |
2593 | ||
f46b5a66 CH |
2594 | long btrfs_ioctl(struct file *file, unsigned int |
2595 | cmd, unsigned long arg) | |
2596 | { | |
2597 | struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root; | |
4bcabaa3 | 2598 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
2599 | |
2600 | switch (cmd) { | |
6cbff00f CH |
2601 | case FS_IOC_GETFLAGS: |
2602 | return btrfs_ioctl_getflags(file, argp); | |
2603 | case FS_IOC_SETFLAGS: | |
2604 | return btrfs_ioctl_setflags(file, argp); | |
2605 | case FS_IOC_GETVERSION: | |
2606 | return btrfs_ioctl_getversion(file, argp); | |
f7039b1d LD |
2607 | case FITRIM: |
2608 | return btrfs_ioctl_fitrim(file, argp); | |
f46b5a66 | 2609 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 2610 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 2611 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 2612 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 2613 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 2614 | return btrfs_ioctl_snap_create(file, argp, 1); |
76dda93c YZ |
2615 | case BTRFS_IOC_SNAP_DESTROY: |
2616 | return btrfs_ioctl_snap_destroy(file, argp); | |
0caa102d LZ |
2617 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
2618 | return btrfs_ioctl_subvol_getflags(file, argp); | |
2619 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
2620 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
2621 | case BTRFS_IOC_DEFAULT_SUBVOL: |
2622 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 2623 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
2624 | return btrfs_ioctl_defrag(file, NULL); |
2625 | case BTRFS_IOC_DEFRAG_RANGE: | |
2626 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 2627 | case BTRFS_IOC_RESIZE: |
4bcabaa3 | 2628 | return btrfs_ioctl_resize(root, argp); |
f46b5a66 | 2629 | case BTRFS_IOC_ADD_DEV: |
4bcabaa3 | 2630 | return btrfs_ioctl_add_dev(root, argp); |
f46b5a66 | 2631 | case BTRFS_IOC_RM_DEV: |
4bcabaa3 | 2632 | return btrfs_ioctl_rm_dev(root, argp); |
475f6387 JS |
2633 | case BTRFS_IOC_FS_INFO: |
2634 | return btrfs_ioctl_fs_info(root, argp); | |
2635 | case BTRFS_IOC_DEV_INFO: | |
2636 | return btrfs_ioctl_dev_info(root, argp); | |
f46b5a66 CH |
2637 | case BTRFS_IOC_BALANCE: |
2638 | return btrfs_balance(root->fs_info->dev_root); | |
2639 | case BTRFS_IOC_CLONE: | |
c5c9cd4d SW |
2640 | return btrfs_ioctl_clone(file, arg, 0, 0, 0); |
2641 | case BTRFS_IOC_CLONE_RANGE: | |
7a865e8a | 2642 | return btrfs_ioctl_clone_range(file, argp); |
f46b5a66 CH |
2643 | case BTRFS_IOC_TRANS_START: |
2644 | return btrfs_ioctl_trans_start(file); | |
2645 | case BTRFS_IOC_TRANS_END: | |
2646 | return btrfs_ioctl_trans_end(file); | |
ac8e9819 CM |
2647 | case BTRFS_IOC_TREE_SEARCH: |
2648 | return btrfs_ioctl_tree_search(file, argp); | |
2649 | case BTRFS_IOC_INO_LOOKUP: | |
2650 | return btrfs_ioctl_ino_lookup(file, argp); | |
1406e432 JB |
2651 | case BTRFS_IOC_SPACE_INFO: |
2652 | return btrfs_ioctl_space_info(root, argp); | |
f46b5a66 CH |
2653 | case BTRFS_IOC_SYNC: |
2654 | btrfs_sync_fs(file->f_dentry->d_sb, 1); | |
2655 | return 0; | |
46204592 SW |
2656 | case BTRFS_IOC_START_SYNC: |
2657 | return btrfs_ioctl_start_sync(file, argp); | |
2658 | case BTRFS_IOC_WAIT_SYNC: | |
2659 | return btrfs_ioctl_wait_sync(file, argp); | |
475f6387 JS |
2660 | case BTRFS_IOC_SCRUB: |
2661 | return btrfs_ioctl_scrub(root, argp); | |
2662 | case BTRFS_IOC_SCRUB_CANCEL: | |
2663 | return btrfs_ioctl_scrub_cancel(root, argp); | |
2664 | case BTRFS_IOC_SCRUB_PROGRESS: | |
2665 | return btrfs_ioctl_scrub_progress(root, argp); | |
f46b5a66 CH |
2666 | } |
2667 | ||
2668 | return -ENOTTY; | |
2669 | } |